DB2 for z/OS: Have You Checked Your Log RBA Value Lately?

I'm in Las Vegas this week for the IBM Information on Demand 2012 conference. At this year's IOD event I've particularly enjoyed learning more about DB2 11 for z/OS, the beta program for which is expected to begin in March of 2013. DB2 11 will deliver a lot in the way of new features and capabilities. One of the big news items here is the extended RBA that will be available in DB2 11 systems running in new-function mode (the RBA, or relative byte address, is an identifier for records written by a DB2 subsystem to its transaction log). Whereas the current RBA is a six-byte hexadecimal value, providing 2^48 bytes (256 TB) of log capacity, the DB2 11 extended RBA is a ten-byte value, allowing for the logging of a yottabyte of data-change records. Not familiar with the term, "yottabyte?" Neither was I until a couple of days ago. It's 1000 zetabytes, or a trillion terabytes. That's a lot of logging capacity.

A larger maximum RBA value is important because a few DB2 for z/OS sites have bumped up against the 2^48-byte limit of the current RBA. Actually, what these folks bump up against is a practical RBA value limit that's a little less than 2^48. See, the RBA value would be thoretically maxed out at x'FFFFFFFFFFFF' (in hexadecimal notation); however, actually reaching that value would prevent a DB2 subsystem from doing much of anything, and that would not be good for data integrity protection. So, what happens is that DB2 will automatically shut down if the log RBA value reaches x'FFFF00000000'. This gives the DB2 subsystem a little room at the end of the log that it will use, when restarted, in resolving any incomplete units of recovery and otherwise ensuring that data is left in a consistent state (an attempt to restart the subsystem in "regular" mode after the automatic shutdown will result in DB2 going through "restart light" processing and shutting down again, after which it could be restarted again with ACCESS(MAINT) for any further clean-up tasks).

That sort of automatic shutdown is of course something you'd want to avoid, and fortunately you can. First of all, DB2 will issue message DSNJ032I, with an "alert level" of WARNING, when the log RBA value reaches x'F00000000000'. This would give you some time to plan for dealing with the approaching RBA limit (DSNJ032I is also issued, with an alert level of CRITICAL, when the RBA value reaches x'FFFF00000000' -- the value that triggers automatic shutdown). Here is a piece of advice for you: do NOT wait until your DB2 subsystem issues DSNJ032I with the WARNING alert level. Instead, check the RBA value for each of your DB2 for z/OS subsystems now. This can be quickly and easily done by issuing the command -DISPLAY LOG on each subsystem. In the command output you'll see a field, H/W RBA, that shows the highest RBA value written by the DB2 subsystem to its transaction log. Chances are, the value that you see in this field will be way lower than x'F00000000000', the WARNING threshold. A colleague of mine who has a lot of knowledge in this area, Paul Bartak, has said that a high-RBA value of x'C00000000000' or greater should be a trigger that gets you into RBA limit planning mode. If what you see for H/W RBA in -DISPLAY LOG output for a subsystem is less than x'C00000000000', just keep an eye on the situation through periodic issuance of -DISPLAY LOG.

If the high RBA value for a DB2 subsystem in your environment is north of x'C00000000000' then as my co-worker Paul suggested you should start developing a plan of action. What you'll end up doing will depend on whether you run DB2 for z/OS in standalone mode or in a data sharing configuration on a Parallel Sysplex. In the standalone case, you will follow a procedure that will result in the DB2 subsystem's RBA value being reset to zero. This procedure is documented in the DB2 for z/OS Administration Guide, under the heading "Resetting the log RBA value in a non-data sharing environment" (you can access the DB2 for z/OS product documentation, including the Administration Guide, in either PDF or HTML form, at http://www-01.ibm.com/support/docview.wss?rs=64&uid=swg27011656). A number of mainframe DB2 sites have successfully gone through this documented RBA-reset process -- it works.

If you have a DB2 subsystem that is approaching the log RBA limit, and that subsystem is a member of a data sharing group, there are a couple of options available to you. One alternative is to reset the member subsystem's RBA -- a less involved process in a data sharing versus a non-data sharing environment. The other choice involves starting up a new member of the data sharing group and permanently quiescing the member with the elevated RBA value (with DB2 10 in new-function mode, one can delete a member from the data sharing group -- until then, the subsystem ID of the permanently quiesced member will continue to show up in the text of some DB2 messages, but that's generally not a big deal). The DB2 for z/OS Data Sharing Planning and Administration Guide, under the heading, "Removing members from the data sharing group," provides information on permanently quiescing a member subsystem. Note that if the new member -- the one that will replace the subsystem that is to be permanently quiesced -- will be the eighth to join the data sharing group, entries in the lock table (part of the lock structure in a coupling facility) will have to be 4 bytes apiece, versus 2 bytes each. To maintain the same number of lock table entries in that case (good for holding the line on false global lock contention), you'd want to increase the size of the lock structure -- which should already be at a power of 2, such as 64 MB -- to the next higher power of 2.

Whether your DB2 subsystem is operating in standalone or data sharing mode, there is information in the DB2 for z/OS Administration Guide, under the heading "Resetting the log RBA," on using output from the print log map utility (aka DSNJU004) to estimate the amount of time you have before a subsystem's RBA value reaches the critical threshold.

Something else for those of you running DB2 in a data sharing group: do NOT think that RBA values no longer matter, just because LRSN values (short for log record sequence number) are used for data recovery operations. RBA values continue to go up for data sharing members as they write records to their log data sets, and the automatic shutdown occurs for a data sharing member just as it does for a non-data sharing subsystem when the RBA value reaches x'FFFF00000000'. I don't want you to be caught napping in this regard.

Speaking of LRSN values, those are also extended beginning with DB2 11 in new-function mode, to the tune of giving you something like another 30,000 years of value growth (LRSN values are derived from the store clock timestamp, and the current LRSN field can take you to the year 2042). Given the 30 years of value-increase that can be accommodated by the current-format LRSN, will any organizations be in a hurry to implement the DB2 11 extended LRSN (done, as for the extended RBA, through a conversion of the bootstrap data set and online REORGs of database objects)? Some will be. Why? Because on occasion a company had to add a delta to the LRSN of the originating member of a DB2 data sharing group in order to have a starting LRSN value that was higher than the already-elevated RBA value of that originating DB2 member. You can look at print log map utility output to see of you have such an LRSN delta in your environment. If you don't -- and that's probably the case -- then the LRSN value limit is still a long ways off for you. If there is a delta, you'll be more interested in getting to the DB2 11 extended LRSN sooner rather than later.

So, to recap: you should periodically check the RBA values of your DB2 subsystems, to ensure that they are well below the critical threshold. If the RBA value for a subsystem is seen to be high (with a 'C' in the high-order position), DON'T PANIC. Prepare to head off the RBA limit situation before it occurs, and get ready to put all this in your rear-view mirror once you get to DB2 11 new-function mode.  

So, Where do You Specify the IP Address of YOUR DB2 for z/OS Subsystem?

If your answer to the question posed in the title of this blog entry is, "In a PORT reservation statement in our z/OS LPAR's TCP/IP profile data set," that's NOT what I want to hear. It's true that for a long time this was the standard way to assign an IP address to a DB2 for z/OS subsystem, and it would look something like this:

PORT 446 TCP DBP1DIST BIND 111.222.333.444  ; DRDA SQL PORT FOR DBP1

This was the way to go because DB2 didn't give you much of a choice in this regard. You can also set things up so that a DB2 subsystem's IP address is specified in a domain name server, but that's not the answer I'm looking for, either. No, if I ask you where the IP address of your DB2 subsystem is specified, what I want to hear is this: "In the bootstrap data set, Mr. Catterall." Of course I'm kidding about the "Mr. Catterall" bit, but the BSDS is indeed where DB2's IP address should be. I'll tell you why momentarily.

Specifying the IP address of a DB2 subsystem in the associated BSDS is something that only recently (relatively speaking) became possible -- the capability was delivered with DB2 9 for z/OS. In a DB2 9 environment, an IP address can be added to a subsystem's communications record in the BSDS by way of the change log inventory utility, aka DSNJU003. The input statement for an execution of DSNJU003 that would place an IP address in the BSDS would be of this form:

DDF IPV4=111.222.333.444

Note that the address of the subsystem could also be specified in the BSDS in IPv6 form (colon hexadecimal). Note, too, that in a data sharing system you can place in the BSDS the IP address of the data sharing group as well as the address of the individual member subsystem. Also in a data data sharing environment it is highly recommended that the address of a DB2 subsystem be a dynamic virtual IP address, or DVIPA (that way, if a failed DB2 member is restarted on a different z/OS LPAR in the Parallel Sysplex, clients utilizing DRDA two-phase commit protocol will be able to re-connect to the member to resolve in-doubt threads).

If you put a DB2 subsystem's IP address in the associated BSDS, a PORT reservation statement for the subsystem in the LPAR's TCP/IP data set would look like this:

PORT 446 TCP DBP1DIST ; DRDA SQL PORT FOR DBP1

That's right: no IP address for the DB2 subsystem in the PORT statement. When a DB2 subsystem's IP address is recorded in that subsystem's BSDS, placing the address as well in a PORT statement for the DB2 subsystem is not only unnecessary, it's ill-advised. Why? Because if the IP address is recorded in both places (the BSDS and the PORT statement), the PORT statement wins. Why is that important? Because if z/OS Communications Server gets a DB2 subsystem's IP address from a PORT reservation statement, you miss out on the benefits of INADDR_ANY functionality. INADDR_ANY is a TCP/IP capability whereby, in a DB2 for z/OS context, a DB2 subsystem can accept requests to its port on any IP address. What that gives you is configuration and operational flexibility. This is particularly advantageous in a DB2 data sharing environment, and when DB2 Connect is running under Linux on System z and utilizing HiperSockets for access to DB2 for z/OS.

There are other goodies associated with recording a DB2 subsystem's IP address in the BSDS instead of in a PORT reservation statement: you don't have to define a domain name for the DB2 subsystem to TCP/IP, and you can use SSL (Secure Socket Layer) for DRDA communications (not possible with BIND specific TCP/IP statements).

You've probably done a lot over the years to modernize your DB2 for z/OS environment. Getting a DB2 subsystem's IP address into the BSDS and out of the PORT reservation statement is another such move. Take this action and position your DB2 subsystem for enhanced client-server application availability, agility, and functionality.

Linux on System z: a Better Place for Your DB2 for z/OS Client-Server Applications

In DB2 for z/OS circles, there has been an evolution of attitudes concerning Linux on System z. Back around 2003, when I was on the user side of the DB2 community, I was part of a team charged with developing a plan for the simplification and streamlining of my company's IT infrastructure. One of the mainframers on this team had a suggestion: why not add a few IFL engines (more on this to come) to our System z servers and use those to consolidate a lot of our Linux systems (which were running primarily on x86 boxes)? I'll confess that this suggestion seemed a little weird to me, a little "out there" -- Linux on a mainframe? Really? The person who was advocating running Linux on System z made some good points, but the idea didn't get far and was set aside. Linux was kind of new to us at the time, and while we were ready to increase the use of this OS in our environment (and we did), we weren't ready to even seriously consider running it on mainframes.

Fast forward to 2010. A few months after rejoining IBM (I'd previously worked for Big Blue from 1982 to 2000), I was working with people at an insurance company where Linux on System z was being put through a trial run. What they were willing to run under Linux on System z was a DB2 Connect gateway, not a more high-profile workload such as an application server or a business analytics tool -- an indication of the iffiness of the Linux-on-z solution as seen by the firm's IT leaders.

Jump now to the present, and it seems to me that a corner has been turned. Over the past couple of years especially, the thinking of many IT people -- both mainframers and Linux specialists -- towards Linux on System has gone from cautious consideration to genuine enthusiasm. All kinds of folks have recently come to the realization that Linux on a mainframe is not some kind of sideways, six-of-one/half-dozen-of-the-other move, in terms of where Linux runs in the enterprise; rather, it's a move that upgrades the IT infrastructure in multiple ways, delivering enhanced performance, cost efficiency, availability, and security. A big chunk of installed System z MIPS, right now, is IFL capacity, and that slice of the pie is growing at a rapid clip (IFLs -- short for Integrated Facility for Linux -- are System z processors that are dedicated to Linux workloads). If I were you, I'd be looking to ride that wave.

In terms of factors that are driving the upsurge in Linux on System z usage, two really stand out: server consolidation and front-ending DB2 for z/OS. That first factor applies to Linux systems that may or may not have any connection to DB2 for z/OS subsystems. A wide variety of Linux workloads can be more cost-effectively run on mainframes versus distributed systems servers. Want to know what one of the big money savers is here? Think software licensing costs: the price of commercial software that runs under Linux is usually based on the number of processors on the system (or the partition) on which the OS is running, and plenty of organizations have found that moving Linux workloads to mainframes results in a decreases in the number of engines needed for those workloads, and that allows for a reduction in licensing costs for vendor-supplied software running in the Linux images.

That reduced-cost advantage also comes into play with regard to the second Linux-on-z driver, DB2 for z/OS front-ending. So do advantages in the areas of performance and security. When a Linux workload -- perhaps an application server running Java programs, or a vendor-supplied ERP or CRM package, or a business analytics tool -- is the client to a DB2 for z/OS server (i.e., if the Linux-based application is a DB2 DRDA requester), that back-and-forth network traffic can flow via HiperSockets. That's memory-to-memory data transfer, folks, and it screams. This "internal LAN" set-up also offers a security benefit, as less of your network is physically exposed to those who might want hack it (as Jim Elliott, one of IBM's prominent Linux-on-z experts, likes to put it, "The best LAN is one with no wires").

At this point, I would be remiss if I didn't bring z/VM into the picture. Can you run Linux on System z without z/VM? Absolutely, but z/VM delivers tremendous virtualization, management, availability, and security capabilities for your Linux-on-z environment. On the security front, RACF for z/VM brings the same peerless validation and auditing features that have made RACF for z/OS famous (note, too, that the cryptographic hardware on System z servers can be exploited for Linux workloads). High availability can be taken higher via the live guest relocation capability of a clustered z/VM configuration, which enables you to move a running Linux image from one z/VM LPAR to another.

Back to dollar (or your local currency) savings: I haven't yet given you the whole picture with regard to the cost-efficiency advantages of Linux on System z. Those IFLs I mentioned are not only great performers, allowing you to do more Linux work with fewer engines -- they are also very attractively priced (and of course they have no impact on the licensing costs for z/OS software running on the same System z server). Want to get even more bang for your bucks? Check out the IBM System z Solution Edition for Enterprise Linux, an offering for existing mainframe systems that combines IFLs, memory, z/VM, and maintenance in a package that's a terrific deal.

Are you wondering who in your shop would administer Linux systems on a mainframe? That's easy: your current Linux system administrators would do the job just fine. During a Linux on System z proof of concept on which I worked a couple of years ago, I recall that a Linux pro in the client's IT department said something to this effect: "It's just Linux." Bingo -- and it's available in distributions from SUSE and Red Hat. What if your organization decides to bring in z/VM with Linux for System z (good move)? Who will take care of that OS, if it's new to your company? Answer: your z/OS systems programmers. First, there's a good chance that one or more of them have some VM experience (many mainframers do). Second, VM has been significantly simplified in recent years (for a "simple" install of z/VM, the instructions take up all of one page).

I'll close by reiterating the fact that Linux on System z is not some Johnny-come-lately solution. It was announced in December of 1999, and it's been regularly enhanced since then. Yes, it took a while for this System z configuration option to catch fire, but it is certainly hot now. If you have Linux systems that connect to DB2 for z/OS, running those Linux images on the same mainframe where the DB2 for z/OS data server runs can take the core of your IT infrastructure to new levels of efficiency, performance, scalability, security, and manageability. Get smart, and check it out.

DB2 10 for z/OS: New Options for Trimming and Upgrading Indexes

First, some explanation as to what I mean by "trimming and upgrading" indexes. My use of "trimming" refers to reducing the number of indexes defined on tables, through elimination of indexes that aren't providing value. Having fewer indexes on a table is good for cost efficiency, as every index increases the CPU cost of every insert and delete operation targeting the table (and every update of an indexed column), increases the cost of execution of several DB2 utilities (including LOAD, REORG, and RUNSTATS), and drives up disk space utilization (though that cost can be mitigated through the index compression capability introduced with DB2 9 for z/OS, about which I blogged a couple of years ago while working as an independent DB2 consultant). That's clear enough, but you may wonder why an index serving no useful purpose got created in your environment in the first place. Perhaps an index created in anticipation that it would be needed for good application performance turned out to be unnecessary for that purpose. Another possibility is that an index that had been necessary for technical reasons became unnecessary (or potentially so) through some enhancement delivered with a recent release of DB2. I'll provide more information on both of these scenarios in the remainder of this blog entry.

The index "upgrade" concept is related to index "trimming," in that identification of an unnecessary index presents you with a couple of options: one is to eliminate the useless index (to reduce CPU and disk costs, as mentioned); the other is to replace that index with one that is useful -- probably in terms of reducing CPU and elapsed time for some important queries (or searched updates or deletes). I'd consider that action to be an index "upgrade" because you've improved application performance while holding the line with respect to the number of indexes defined on a table (one useless index out, one useful index in).

Whether your aim is index trimming, index upgrading, or both, the essential first step is identification of indexes that can be eliminated because they no longer serve a useful purpose (if indeed they ever did). The last several releases of DB2 for z/OS have made this task easier and more fruitful. I'll first provide a quick review of enhancements related to index trimming and upgrading that were introduced with DB2 V8 and V9, then I'll get into new opportunities for tightening up and/or boosting the effectiveness of your index configuration that were made available through DB2 10 for z/OS. So, the review:

• Table-controlled table partitioning (DB2 V8). Though this enhancement was delivered years ago, a lot of mainframe DB2 shops have yet to leverage its value (as I pointed out in an entry I posted to this blog a few months ago). Basically, it comes down to this: when you convert an index-controlled partitioned table space to an table-controlled partitioned table space, the formerly partition-controlling index may end up becoming useless. There's your opportunity for index trimming or upgrading.
• The LASTUSED column of the SYSIBM.SYSINDEXSPACESTATS catalog table (DB2 9). Those "I thought we'd need it, but maybe we don't" indexes were once pretty hard to find. Looking for package dependencies on an index will tell you when one is not used by static SQL statements, but what about dynamic SQL? The LASTUSED column of the real-time statistics table SYSIBM.SYSINDEXSPACESTATS makes identification of indexes that are just taking up space much easier than before, and that multiplies your index trimming and upgrading opportunities. I've blogged about this DB2 feature several times, most recently in an entry posted to this blog a couple of years ago.
• Index-on-expression (DB2 9). Index upgrading is all about replacing an unnecessary index with one that does you some good. DB2 9 broadened the "does you some good" landscape by introducing a new category of indexes, that being indexes defined not just on columns, but on column expressions, such as SUBSTR(DEPT_NAME, 2, 3). With such an index, a predicate that had been non-indexable can be made indexable, leading to potentially dramatic query performance improvement. I blogged about this DB2 enhancement a few years ago.
• The XML data type (DB2 9). Just storing XML data in a column defined with the XML data type (as opposed to, for example, the VARCHAR or CLOB data type) can deliver greatly improved performance with respect to querying said XML data. Create an index on that XML column (via XPath expressions allowed in index definitions), and you can REALLY speed things up (particularly when the XML documents are pretty large). This is another of those index upgrade opportunities -- and keep in mind that it pertains not only to retrieval of XML data, but to retrieval of non-XML data based on a predicate that references an XML column (an excellent reference for XML data in a mainframe DB2 environment is the IBM "red book" titled Extremely pureXML in DB2 10 for z/OS -- most of the information in which is also applicable to a DB2 9 system).

OK, now for the good stuff delivered through DB2 10 for z/OS that can take your index trimming and upgrading efforts to new levels of effectiveness. I'll start with what are called index INCLUDE columns. This is really big, in that it created a whole new category of indexes that could be eliminated, thereby paving the way for new index trimming and upgrading actions. The concept of index INCLUDE columns is simple, and best explained through an example. Suppose you have a table for which the combined values of columns C1 and C2 in each row have to be unique. To enable DB2 to enforce that requirement, you defined a unique constraint on (C1, C2), and created a unique index on (C1, C2). Suppose further that you wanted to provide index-only access for queries that reference only columns C1, C2, C3, and C4 of the table, or that you wanted to enable sort avoidance for queries that contain ORDER BY C1, C2, C3, C4. In a pre-DB2 10 environment, you addressed this second need by creating an additional index on columns C1, C2, C3, and C4. DB2 10 (in new-function mode) provides a new option: you can alter the unique index on (C1, C2) to INCLUDE (C3), and again to INCLUDE (C4) (for ALTER INDEX, only one column can be named in an INCLUDE specification -- n columns are included in an index through n executions of ALTER INDEX). In doing that, you'd end up with one index that could accomplish two purposes: 1) enforcement of uniqueness for (C1, C2) value pairs, and 2) query performance enhancement (through enablement of index-only access for queries referencing only columns C1, C2, C3, and C4; and sort avoidance for queries containing ORDER BY C1, C2, C3, C4 ). Voila! That index on (C1, C2, C3, C4) is now unnecessary and can be eliminated, clearing the way for index trimming or index upgrading. Note that INCLUDE is an option for CREATE INDEX as well as ALTER INDEX statements (in the case of CREATE INDEX, several columns can be listed in the INCLUDE part of the statement, so INCLUDE (C3, C4) is OK, syntax-wise). Whether it's an ALTER INDEX or a CREATE INDEX statement, the aim is the same: use one column (or set of columns) for enforcement of a UNIQUE constraint (or for a primary key), and with the same index include another column (or set of columns) that are "along for the ride" to provide better-performing query access paths. The extra columns added to an index via INCLUDE do NOT factor into uniqueness determination with regard to the "core" key on which the unique index is defined -- that's what enables dual-purpose indexing.

Here's another index trimming/upgrading opportunity provided courtesy of DB2 10: hash-organized tables. This is a feature that has generated a good bit of buzz in the DB2 community, but some folks have overlooked the tie-in to index trimming and upgrading. Here's how hash organization of data in a table works: when a table is created (in a DB2 10 new-function mode environment) with ORGANIZE BY HASH (or when an existing table is changed via ALTER TABLE to ADD ORGANIZE BY HASH), assignment of rows to table pages is not based on a clustering key; rather, the hash key value for a row (this is a unique key specified by the user) is run through a hashing algorithm whose output determines the target page. Generally speaking, when people think of hash-organized tables, they think of the performance boost that can be achieved when a query referencing the hash key in an equals predicate can get a row from DB2 with a single GETPAGE. That's great (for single-row access -- not so much for sequential data access), but here's something else to consider: if you're looking to alter an existing table to be hash-organized, and the unique key that you're going to use for hash-organization purposes is already defined as unique and has a corresponding uniqueness-enforcing index (and this applies as well to a primary key), that unique index will likely be unnecessary following the change to hash organization. Why? Because DB2 does not need an index on the hash key to ensure uniqueness of that key -- it can accomplish duplicate-value checking very efficiently using just the table (remember that the location of a row in a hash-organized table can be directly ascertained using the value of the hash key in the row). So, you can probably drop the index that had been previously created on the key that is the hash key for the hash-organized table, and there you have that new opportunity for index trimming or upgrading. Now, I say that you can probably drop that unique index, because it's possible that you'd want to keep it for queries that 1) reference only the column or columns in that index and 2) involve a scan of index values (recall that hash-organized tables are not good for the performance of data retrieval operations that involve sequential access).

[Note: technically speaking, converting a table to be hash-organized results in the creation by DB2 of a new index on the table, but this index -- the hash-overflow index, used to locate rows that were stored in the table space's overflow area because the hash-determined target page was full -- is a "sparse" index, containing entries only for rows in the overflow area. It does not have the CPU and disk space costs associated with traditional indexes.]

So, index trimming and upgrading is good for the efficiency of your DB2 for z/OS environment, and good for the performance of your DB2-accessing applications. DB2 10 continues the trend of giving you more opportunities to prune your index configuration, or to maintain a certain level, with respect to the number of indexes defined, while boosting overall index effectiveness (through replacement of unneeded indexes with value-delivering indexes). Invest some time in this area, and you're likely to see a good return.

The New IBM zEnterprise EC12: Big Iron Gets Bigger (and Better)

When I say "bigger," I don't mean bigger in a "footprint" sense -- the new flagship system in IBM's mainframe server product line is very close, size-wise, to its predecessor, the z196, and has similar power draw and heat load characteristics. What I mean by "bigger"is that the zEnterprise EC12 provides the ability to get more work done, in less time, more efficiently than ever before. That's good news for organizations that need best-of-breed enterprise computing capabilities. It's also good news for those of us who comprise the DB2 for z/OS community: the horse that our favorite DBMS rides is, in the form of the EC12, a thoroughbred like no other.

Continuing with the equine analogy, let's start with a look at the raw horsepower numbers. The EC12's engines run at 5.5 GHz, the highest clock frequency in the industry (the frequency of a z196 engine is 5.2 GHz). That's impressive enough, but the EC12 delivers a per-core processing capacity advantage over the z196 -- up to 25% -- that is significantly greater than suggested just by the difference in engine speed. Very fast processors are only part of the EC12 performance story. There's also a doubling of the cache memory on each processor chip, which reduces delays caused by having to go "off chip" for instructions and data.

Not only are there faster engines that can be more effectively kept busy, there are more engines: a single EC12 server can have up to 120 cores, up to 101 of which can be configured for client use (versus a maximum of 80 configurable cores on a z196 server). Put it all together (a larger number of faster engines), and a fully configured EC12 can deliver 50% more processing capacity than a top-end z196 server.

The total amount of memory available on an EC12 server -- 3 TB -- is the same as for the z196, but something new and very cool is being done with EC12 memory: it can be managed in 2 GB page frames. Yes, that is 2 GB, as in 2000 times the size of the 1 MB page frames available on z196 servers. What does that mean for DB2? Think about a page-fixed buffer pool of 2 GB in size (e.g., 500,000 4K buffers) fitting into ONE real storage page frame. Virtual-to-real storage translation will be more efficient than ever, and that will translate into CPU savings. If you've ever wondered, as a mainframe DB2 person, whether or not the age of Big Memory has really begun, wonder no more. It's here. Get ready to exploit it, if you haven't already.

You might find yourself thinking, "What would my company do with a server that has that much memory and that much processing capacity?" In that case, I have one word for you: virtualization. Don't limit your thinking of EC12 exploitation to one or just a few system images. Think LOTS of system images. Lots of z/OS LPARs on an EC12? Maybe, but I'm thinking more along Linux lines when I think of lots of virtual systems running on a single EC12 box. If your organization isn't running Linux on System z, or thinking seriously about doing so, you're missing a train on which a whole lot of people are traveling. Installation of IFL engines on zEnterprise servers (these are specialty engines that run the Linux operating system) is soaring, and with good reason. The mainframe is a great virtual-system platform (has been for decades, since long before "cloud" mean anything other than coalesced water vapor), and if you have a big and growing DB2 for z/OS client-server workload (as many organizations do), what better place is there for your DB2-accessing application servers than in a Linux image on the same System z server in which DB2 is running? Ever heard of HiperSockets? Network performance doesn't get any better than that. And, think of the benefits in terms of network simplification (and security) when app servers run under Linux on System z, instead of on outboard boxes. With the EC12, the already great System z virtualization story gets better still.

Virtualization's hot, and so is analytics. How about analytics on z? I'm not talking just about having the data accessed by your analytics tools housed in a mainframe database -- we already know that DB2 for z/OS is a great solution there. I'm talking about having the analytics tools themselves running on System z -- in a Linux image or a z/OS LPAR. More and more organizations are doing just that, and the EC12 will provide a related momentum boost. How's that? Well, as you get into progressively higher-value types of analytics -- from "what happened" reporting to "what will happen" predictive modeling -- then you find that the associated processing gets to be more and more CPU-intensive. The EC12 delivers here with significant improvements in performance for compute-intensive and floating-point applications. C and C++ on z? Java on z? Yes. The EC12 is the best mainframe server yet for business analytics.

Excellent performance across a wide range of applications is great, but we all know that a system has to be up if it's to deliver the processing power your organization needs. System z has always been the high-availability champ, and the EC12 takes high availability to a whole new level with zAware (short for System z Advanced Workload Analysis Reporter). zAware provides what one of my colleagues has termed "integrated expert system diagnostics," constantly monitoring OPERLOG messages (which at some sites can number in the millions per day) and presenting related information via an easy-to-interpret graphical user interface. zAware can help mainframe operations personnel to quickly determine when an EC12 system is not behaving normally -- thereby enabling corrective action to be taken before a situation reaches user-impacting proportions.

I've covered a lot in this blog post, but you can learn more about the zEnterprise EC12 and I encourage you to do so. Use the hyperlink I provided at the top of this entry, check out the IBM EC12 Technical Introduction and EC12 Technical Guide "red books," and look for presentations at national conferences and local events. The more you know, the more you'll realize that Big Iron never looked better.

When CICS and DB2 for z/OS Monitors Disagree (Part 2)

About a week ago, in part 1 of this two-part blog entry, I described a situation I've encountered several times at mainframe DB2 sites over the past 20 years: CICS users complain of elongated response times for DB2-accessing transactions, and these users' complaints are backed up by CICS monitor data that indicate elevated "wait for DB2" times as being the cause of the performance degradation. Concurrently, DB2 systems people point to their own data, from a DB2 monitor, showing that in-DB2 time for CICS transactions is NOT significantly higher than normal during times when the CICS programs are running long. Finger-pointing can ensue, and frustration can build in the face of a seemingly intractable problem: how can it be that the CICS monitor shows higher "wait for DB2" times, while according to DB2 monitor data the performance picture looks good? I noted in the aforementioned part 1 blog entry that this situation can occur when there is either a shortage of DB2 threads available for CICS-DB2 transactions or a shortage of TCBs through which the DB2 threads are utilized. In this part 2 entry I'll shine a light on another possibility: the disagreement between CICS and DB2 monitors can be caused by inappropriate priority specifications for DB2 address spaces and/or CICS transactions.

The priority to which I'm referring is dispatching priority -- the kind that determines which of several "in and ready" tasks in a z/OS system will get CPU cycles first. Dispatching priorities for address spaces are specified in the WLM policy of the z/OS LPAR in which they run. The busier a system is (with respect to CPU utilization), the more important it is to get the address space dispatching priorities right. DB2's IRLM address space (the lock manager) should have a very high priority, and in fact should be assigned to the SYSSTC service class (for ultra-high-priority address spaces). Most sites get this right. The mistake that I've seen more than once is giving the DB2 system services and database services address spaces (aka MSTR and DBM1) a priority that is lower than that of the CICS application owning regions (AORs) in the z/OS LPAR. The thinking here is often something like this: "If the DB2 MSTR and DBM1 address spaces have a higher priority than the CICS AORs, they will get in the way of CICS transactions and performance will be the worse for it." Not so. The DB2 "system" address spaces generally consume little in the way of CPU time (DDF can be a different animal in this regard, as I pointed out in an entry posted to this blog last month), so "getting in the way" of CICS regions is not an issue. In fact, it's when CICS regions "get in the way" of the DB2 system address spaces (as they can in a busy system when they have a higher-than-DB2 priority), that CICS transaction performance can go downhill.

How's that? Well, it can come down to a thread-availability issue. It's common for CICS-DB2 threads to be created and terminated with great frequency as transactions start and complete. The DB2 address space that handles thread creation and termination is MSTR, the system services address space. If CICS address spaces are ahead of MSTR in the line for CPU cycles (i.e., if the CICS AORs have a higher priority than MSTR), and if the z/OS LPAR is really busy (think CPU utilization north of 90%), MSTR may not be readily dispatched when it has work to do -- and the work MSTR needs to do may be related to CICS-DB2 thread creation. If DB2 threads aren't made available to CICS transactions in a timely manner, the transactions will take longer to complete, and "wait for DB2" will be seen -- via a CICS monitor -- as the reason for the not-so-good performance. Check DB2 monitor data at such times and you'll likely see that things look fine. This is because DB2 doesn't "see" a CICS transaction until it gets a thread. Thus, as I pointed out in part 1 of this two-part entry, the "wait for DB2" time reported by the CICS monitor can be time spent in-between CICS and DB2, and that's why the different monitors paint different pictures of the same performance scene: the CICS monitor indicates that transactions are waiting longer for DB2, and that's true, but the DB2 monitor shows that when a transaction does get the thread it needs it performs just fine. If you see this kind of situation in your environment, check the dispatching priorities of the DB2 address spaces and of the CICS AORs. The priority of the DB2 MSTR, DBM1, and DDF address spaces should be a little higher than that of the CICS AORs in the LPAR. [Don't sweat giving DDF a relatively high dispatching priority. The vast majority of DDF CPU utilization is associated with the execution of SQL statements that come from network-attached clients, and these execute at the priority of the application processes that issue the SQL statements -- not at the priority of DDF.] Oh, and in addition to ensuring that DB2 MSTR can create threads quickly when they are needed for CICS transaction execution, think about taking some of the thread creation pressure off of MSTR by increasing the rate at which CICS-DB2 threads are reused -- I blogged on this topic a few months ago.

One other thing to check is the priority of the CICS transactions themselves -- more precisely, the priority of the TCBs used for CICS transaction execution. This is specified through the value given to PRIORITY in the definition of a CICS DB2ENTRY resource, or in the DB2CONN definition for transactions that utilize pool threads. The default value of the PRIORITY attribute is HIGH, and this means that the tasks associated with entry threads (or pool threads, as the case may be) will have a dispatching priority that is a little higher than that of the CICS AOR's main task. HIGH is an OK specification if the z/OS LPAR isn't too busy -- it helps to get transactions through the system quickly; however, if the z/OS LPAR is very busy, PRIORITY(HIGH) may lead to a throughput issue -- this because not only the DB2 address spaces, but the CICS AORs, as well, could end up waiting behind transactions to get dispatched. In that case, going with PRIORITY(LOW) could actually improve CICS-DB2 transaction throughput. I have seen this myself. Bear in mind that PRIORITY(LOW) doesn't mean batch low -- it means that the transactions will have a priority that is a little lower than that of the CICS region's main task.

Bottom line: dealing with (or better, preventing) CICS-DB2 performance problems is sometimes just a matter of getting your priorities in order.

When CICS and DB2 for z/OS Monitors Disagree (Part 1)

Today I'm writing about a situation that I first encountered almost 20 years ago, and which I most recently saw about three months ago: an organization with a CICS transaction workload accessing a DB2 for z/OS database reports a performance problem. The company's CICS people point to monitor data that indicate higher "wait for DB2" times when CICS transactions run long. Meanwhile, the folks on the DB2 team serve up monitor numbers of their own, showing that in-DB2 times for CICS transactions are not higher than normal during CICS slow-down events. What's to be made of this seeming contradiction? How can it be that for a given period, a CICS monitor shows unusually high "wait for DB2" times while DB2 monitor data for the same period show consistently good in-DB2 elapsed times for CICS transactions? I've found that this kind of "are too!" / "am not!" situation is usually the result of CICS transactions waiting longer than usual for DB2 threads. "Wait for thread" goes in the "wait for DB2" bucket from the CICS monitor perspective, but it doesn't inflate in-DB2 elapsed times because the DB2 monitor doesn't "see" a transaction until it has a thread. "Wait for thread" time is, in essence, a waiting that is "between" CICS and DB2, and that's why the CICS and DB2 monitors see it differently.

OK, so why would CICS transactions have to wait longer than usual for DB2 threads? In my experience, this happens for one of two reasons: not enough threads, or not enough priority. In the remainder of this blog entry I'll expand on the first of these two factors: not enough threads (in some cases, this is actually a matter of not having enough TCBs, as explained below). In a part 2 entry, I'll address the priority issue.

Back in the 1980s and 1990s, the connection between a CICS application-owning region (AOR) and a local DB2 for z/OS subsystem was defined by way of a macro called the RCT, or resource control table. RCT gave way to RDO (CICS resource definition online) towards the end of the 1990s. One of the CICS-DB2 connection values, specified in the definition of a resource called DB2CONN, is TCBLIMIT -- the maximum number of TCBs (task control blocks) that can be used to connect transactions running in the CICS region to a target DB2 subsystem. Another of the CICS-DB2 connection set-up parameters, THREADLIMIT, appears in a DB2CONN resource definition (indicating the maximum number of pool threads for CICS-DB2 transactions) and can also appear in a DB2ENTRY resource definition (indicating the maximum number of entry threads for transactions associated with the DB2ENTRY resource). The sum of all THREADLIMIT values (for pool threads and entry threads) for a given CICS region should be less than the value of TCBLIMIT for the region, and people generally start out that way; however, over time folks may increase THREADLIMIT values -- to accommodate a growing CICS-DB2 transaction workload -- without adjusting TCBLIMIT accordingly; thus, the sum of all THREADLIMIT values for a CICS region could end up being greater than the TCBLIMIT value, and that could result in a situation in which threads are available for transactions, but TCBs needed to use those threads to connect to DB2 are insufficient in number to avoid elongated wait times. You should check your system for this possibility, and, if the sum of THREADLIMIT values exceeds TCBLIMIT for a region, either adjust TCBLIMIT upwards or adjust THREADLIMIT values downwards. I'd generally lean towards a larger TCBLIMIT value in that case, but if there were a lot of protected entry threads defined (PROTECTNUM > 0 and THREADLIMIT > 0 for DB2ENTRY resources), I'd consider reducing PROTECTNUM and THREADLIMIT values for those DB2ENTRY resources.

Here's another scenario: TCBLIMIT is greater than the sum of all THREADLIMIT values for a CICS region, but wait-for-thread time is still relatively high. This can happen when the following are true:

• THREADLIMIT is set to zero (or a very small number) for DB2ENTRY resources
• THREADWAIT(POOL) is specified for these same DB2ENTRY resources
• THREADLIMIT is a too-small number for pool threads (i.e., THREADLIMIT has a small value in the region's DB2CONN resource definition)

In that case, it may be that lots of transactions are overflowing to the pool, but the number of pool threads is not large enough for the associated transaction volume. This situation could be indicated by a non-zero value in the W/P column for the POOL row in the output of a DSNC DISPLAY STATISTICS command for the region in question (this is an attachment command issued through a CICS-supplied transaction called DSNC). If you see such a value, bump up the THREADLIMIT number in the DB2CONN resource definition for the region (and in doing that, check, as previously mentioned, to see if a TCBLIMIT increase might be needed, as well).

Of course, you could also have elongated wait-for-thread times if a DB2ENTRY resource definition has a very small (but non-zero) THREADLIMIT value and a specification of THREADWAIT(YES). Before messing with that, check to see if this is an intentional aspect of the DB2ENTRY definition: it's conceivable that a certain transaction must be limited with respect to concurrent execution (maybe even single-threaded, via THREADLIMIT(1) and THREADWAIT(YES) for a DB2ENTRY resource) in order to prevent a contention problem.

Here's one more possibility (though it's not something that I've actually seen): you may have high CICS wait-for-thread times because the DB2 limit on the number of threads for local (i.e., not network-attached) programs is too small. The ZPARM parameter CTHREAD specifies this limit, and it's for all local threads: for CICS transactions, batch jobs, TSO users, etc. If you have an indication of high CICS wait-for-thread times, and you have a pretty small CTHREAD value, consider adjusting CTHREAD upwards. Note that CTHREAD can have a much larger value in a DB2 10 environment versus prior versions of DB2 for z/OS -- this because almost all thread-related virtual storage usage goes above the 2 GB "bar" when packages are bound (or rebound) in a DB2 10 system. Whereas the sum of CTHREAD and MAXDBAT formerly had to be less than 2000, in a DB2 10 environment that sum has to be less than 20,000.

You can help to avoid high CICS wait-for-thread times by ensuring that you have enough threads (and enough CICS TCBs) for your CICS-DB2 transactions. Check back in a week or so for part 2 of this two-part entry, in which I'll explain how high wait-for-thread times can be a matter of priorities (dispatching, that is).