02/16/2011
I recently received a question regarding journal caching from a reader. His concerns led me to believe that it might be helpful to share with others the answer I drafted for him. His question went something like this:
“I’ve found your Journal insight articles at System iNetwork quite educational. In addition, I’ve read the IBM “Journal Caching: Understanding the Risk of Data Loss” tech note and the “Striving for Optimal Journal Performance” Redbook at the IBM Redbooks site.
"However,
02/02/2011
One of the nice things about working in the lab is that you get the opportunity to play detective. Customers will often show up, application and files in hand, asking for help in figuring out what’s going on and what they can do about it.
One of the more interesting (and challenging) instances I've encountered involved a shop that was planning for future growth of their business while trying to decide how to accommodate it. They had taken an instance of their daily activity, ramped it up to
02/01/2011
Write cache can be an impressive tool for improving the performance of applications that aggressively update journal-protected database files.
01/05/2011
As journal developers, our objective in the laboratory was always to try to stay two steps ahead of the folks who would be using the operating system software we were developing. That meant we had to look into our crystal ball and try to anticipate where folks would be 24 months or so in the future.
One of the most challenging areas to anticipate turned out to be maximum capacity limits. IBM often referred to these as "limits to growth," and we even had an employee in the lab who attempted to
01/01/2011
Simplifying Filename Lists
Q. I am using the following to get a list of files in a directory:
OVRDBF FILE(STDOUT) TOFILE QTEMP/DIROUTPUT)
CHGVAR VAR(&QSHCMD) VALUE('find' *BCAT '/edi/inbound' +
*BCAT '-type f')
QSH CMD(&QSHCMD)
Is there a way to limit the above command to just the inbound directory? We have an archive subdirectory below inbound, and it is pulling these files in, as well. Also, can I just get the filenames instead of the path plus filenames?
A. This command
12/15/2010
You've been diligent by enabling journal protection for your critical files; you've been rigorous about saving those files to tape each night; you've been fastidious about assuring that those tapes get shipped off site to a safe storage location promptly; and you've left no stone unturned in your efforts to similarly assure that your journal receivers are saved in a timely fashion, as well. You're feeling pretty good, right?
What more is there to think about?
Answer: How long is the actual
12/01/2010
Do you make use of data queues in your shop? Do they house information you care about? Does recovering the contents of these objects in the event of a disaster matter? How well prepared to do so are you? Would you earn a gold star?
If you're using the standard defaults when you save your data queues to tape, you may be surprised to learn how little you are actually saving.
A Little History
A few releases ago, the journal development team in Rochester was asked to lend a hand to
12/01/2010
These free journal tools from IBM are some of the same tools the gurus in the Rochester lab have used for years—and they can help you analyze like a journal guru.
11/17/2010
My dad owned and operated a restaurant. In fact, my earliest job was as a dishwasher in that restaurant. The health department in our county was well known as one of the toughest in the whole state. It gave my dad particular pleasure to note that the local health inspectors stopped in for lunch nearly every day. They had confidence that the kitchen was spotless.
It similarly gives journal developers in the IBM laboratory pleasure to note that certain portions of the rest of the operating
11/03/2010
You may be among the hundreds of shops that continue to use database practices that sounded good at the time but have become long in the tooth. If you have journal protection enabled for your database files, some of these practices may be costing you dearly and thwarting the ability of the journal to build up a head of steam. Maybe it's time to consider shedding some outdated practices.
A Little History
Early in my career I worked on the microcode for the underlying database design that,
11/01/2010
As new high availability methods arrive for IBM i, journaling remains key for catching critical data.
10/20/2010
It's a question some astute journal users wrestle with: Should I continue to collect so-called BEFORE images within my journal receiver, or should I toss them?
There are pros and cons to both approaches. There's also a new release 7.1 feature of IBM i that can make the decision a bit easier for some remote journal users.
Some Basics
Let's start with some basics. Are BEFORE images essential? No, not for most operations you're likely to perform. IPL recovery, for example, doesn't mandate
10/06/2010
Even folks who suspect they aren’t journal users are likely to be - - which may come as somewhat of a surprise. That’s because there’s a subtle variety of journaling which kicks in behind the scenes if the operating system suspects that your database keyed access paths are bigger than a breadbox.
This normally clandestine form of journaling often uses hidden journal receivers.
Not all access paths (you may know them better as SQL indexes and/or Keyed Logical Files) have such clandestine
09/15/2010
No, we’re not talking about middle-age spread. Rather we’re talking about the tendency on IBM i for a Journal receiver to spread itself across a set of disk drives. This isn’t a haphazard consequence of the fact that IBM i employs an addressing technique known as single-level-store but rather a deliberate positioning technique which the journal receiver (and very few other objects on IBM i) deliberately employs. For you see, the default behavior on IBM i with single level store is that each
09/01/2010
Have you got certain housekeeping chores in your life that you feel compelled to perform time and time again?
For me, it’s dirty floor mats in my car. There’s just something about grit and grime being ground into my carpet that I find particularly distasteful. As a consequence, I carry a portable hand-held vacuum in my back seat. While I’m not so fastidious that I reach for the vacuum each time I exit the vehicle, I have been known (especially during Minnesota winters with lots of sand on
08/18/2010
Each spring, round about the middle of the month of May, I’d place a sign on my desk. It read “Gone Fishing”. It announced to my IBM colleagues that an important rite of spring had greeted Minnesotans and that along with hundreds of thousands of other hardy Minnesota residents, I had hitched up the boat, grabbed the chest waders, and was on my way north to find the elusive Walleye - - the state fish. The sign also kept my IBM colleagues abreast of the fact that it would do them little good
08/04/2010
Due to improved hardware reliability, experiencing a so-called abnormal IPL (one which follows a crash) is becoming less and less common - - and that’s a good thing. However, as such experiences become less prevalent; one can get lured into taking their focus off the kinds of practices which can help tamp down the duration of such abnormal IPLs when they do ensue.
It’s somewhat like my experience with changing flat tires. Knock-on-wood, but as best I can recall it’s been nearly a decade
07/21/2010
Sometimes the normal tools you keep in your tool belt are simply inadequate to deal with the job at hand. In those instances you need to up the ante and get your hands on a tool capable of dealing with the challenge that lies ahead.
Commitment control users can feel that way if they’re faced with a long-running transaction that just doesn’t seem to end.
A few years ago, phone calls of this nature came into the lab about once a month. Often the frantic voice on the other end of the line
07/07/2010
A bloated journal receiver is no fun. It fills rapidly, it takes extra time and space to save to tape, it chews up disk space and, if you happen to also be a remote journal user, it clogs the TCP/IP communication line.
Hence, it makes sense to use all the tools in your tool-belt in an effort to help keep your journal receivers modest in size.
One of the underutilized means of doing so is as simple as enabling the optional Journal Minimal Data attribute.
This is an especially attractive
07/01/2010
IBM i 7.1 contains several improvements targeted at high availability users who employ remote-journal support. Use these new features to enhance your journal behavior, particularly in an HA environment.