As there are no official tool or instructions for doing this, I thought I would take a moment to give some brief instructions on how you can achieve this yourself.

At the moment, in WMB 8 and IIB 9 only two tables are used to store recorded data: “wmb_msgs” stores each recorded message and some metadata, and “wmb_binary_data” is used if a message body or exception list was recorded. As the tables have no enforced referential integrity all we have to do when deleting data is remove it from the tables in the correct order. For example you could use the following commands to remove messages older than 15th November 2011

DELETE FROM wmb_binary_data WHERE wmb_msgkey IN (SELECT wmb_msgs.wmb_msgkey FROM wmb_msgs WHERE event_timestamp < '2011-11-15 14:54:35.121')
DELETE FROM wmb_msgs WHERE event_timestamp < '2011-11-15 14:54:35.121'

As you will likely want a more complicated WHERE clause, and to check it carefully, you may want to create a sql instruction file, for example with the following content:

-- connect to DB
CONNECT TO rrrdb
-- set an alternate schema if the default doesn't apply to you
-- SET SCHEMA WMB
-- now do our deletes
DELETE FROM wmb_binary_data WHERE wmb_msgkey IN (SELECT wmb_msgs.wmb_msgkey FROM wmb_msgs WHERE event_timestamp < '2011-11-15 14:54:35.121')
DELETE FROM wmb_msgs WHERE event_timestamp < '2011-11-15 14:54:35.121'

Save this in a file, and you can then edit it and check it carefully, then run it with db2 -f purge.sql

You will notice that we are using strings as our timestamps. This is actually correct, as they are stored as VARCHAR in the table as we couldn’t think of any reason to parse then and turn then into timestamp types in the database, so instead went for the slightly better performing option of a VARCHAR. This of course means you must format your dates in the full ISO standard, with YYYY-MM-DD, always using two digit months and days.

You can also add your own additions to the where clause, just make sure you add it to both DELETE statements.

You don’t have to do this on a db2 command line of course, you could write your own script or application to run the database commands for you, and even add it to a cronjob if desired.

N.B. I have only tested this with db2, though I am pretty certain the same SQL should work in any other database supported by Record & Replay.

(Also posted on the official IBM Integration Bus blog)

The requirements were pretty simple. It had to house a mini-ITX motherboard, two hard drives, with space to expand for four or six if I ever require it (though powering them may be another matter), and be reasonably accessible. I had decided that the computer didn’t need to be visible from the front, so started thinking about a taller, vertical case to drop down behind the TV. This seemed like a better idea, than any desktop style case sitting under the TV.

As it was never going to be seen, I simply bought a large piece of MDF, and started by simply drawubg outlines for a few components I wanted space. I decided to leave space for a full-sized dvd drive in case I wanted to add it later. For the hard drives, I already had an old drive cage from another PC. The mini-ITX motherboard layout can be found online and so I simply measured, rather carefully, where to put the mounting risers and drilled holes for them. To my surprise they screwed in and held without needing any glue.

Marking out the components and putting in the motherboard mounts

I had decided by this point that all connectors for the PC were going to be at the top of the case, so I had to cut a large hole for all the IO connectors on the motherboard, and also drilled some holes for power and reset switches. Having just guessed a random size for the ends, I joined the end pieces with dowel rods to add some strength.

The power connector for the motherboard was an angled dc barrel plug, so that needed some extra carving of space to allow it to fit fully in. At this point I was positioning the motherboard in place to check the IO connectors lining up, and was rather amazed that all the mounting holes were perfectly in place

Lining up the side panel with the rest of the case, and drilling holes for reinforcement dowel rods

Starting to look like a pc case

Once I had enough side pieces cut and the sizes worked out, I could glue the first pieces in place permanently.

The bottom was going to contain an enormous (and rather quiet) fan, so I had to cut a hole for it. I didn’t own a jigsaw, so had to wait while I found a nice cheap one online. It turns out I can cut a pretty neat circle with a jigsaw, and the help of a compass to trace the circle. A dremel was helpful to neaten it up too.

Some larger dowel rods formed feet to gives some space for airflow

The lid was to be formed of two pieces of wood together, so that after opening I would have access from one side and the front. I also drilled a load of air holes in the edges to allow the hot air out of the case. You can also see the wooden power and reset button covers, and holes for power and hdd led lights.

The hard drive cage now mounted. A layer of silicone sealant underneath to reduce vibration, and some screws to prevent it going anywhere. I also added an aluminium L-shaped bar to provide some strength to the bottom piece.

Simple hinges held the lid on – these took time to cut the grooves for. Glad I had a dremel, as it made it a lot faster

The catch to hold the lid closed. I looked at lots of styles, and this seemed the best, though I did consider a magnetic clasp

Close-up of the top of the case, showing the power buttons and an extra hole for a serial port

The feet fitted in rather snugly, but my attempts at silicone sealant soles to reduce noise failed

Mounting the components. The fan was simply stuck with silicone sealant, and the hard drives only had screws on one side

A leather handle for lifting and it is time to wire it all up

All the components in the finished case

Afterthoughts: It turned out to be quite easy to just place things on wood, trace them and put it together, but I’m not sure it is the best way to build something like this. I should probably have modelled something first, maybe digitally, maybe just out of card. The hardest thing was that having started putting it together I realised that the lid was never going to open fully, as the width of the wood got in the way of the opening. Having the lid contain three sides could have helped here, but at the time I didn’t think I would need an aluminium bar to add strength, but didn’t think it would have enough strength for just one end piece to support itself. If I had thought of the aluminium reinforcement before glueing the side panel on then I could have done it differently.

Most of the components on the top of the case, the power buttons, leds, serial connector and IO plate were simply glued in place with a hot glue gun. This worked well, but was a bit messy.

I would have loved to have a go at a metal case, but I simply don’t have any tools for bending metal, so it wasn’t going to be practical. Also, wood is much easier to cut.

MDF seems to have been a poor choice of material. It often split when drilling into the end of the wood, such as drilling holes for the dowel or screwing in the aluminium support piece. Chipboard or plyboard is probably a better choice as it has a bit more strength, or better yet – real solid wood!

I ended up with a lot of space in the case, which is great for working on it, but having only two accessible sides limits access quite a lot. It would have been worth thinking about having the lid differently so that access to the front and two sides was possible. This could also allow it to be smaller but still have good access.

Avoiding hard drive vibrations is tricky. The sealant helps I think, but I would have liked to have had the time to mount the whole thing on rubber grommets, as it does sit in the living room. It is still quieter than the electronics for backlighting the LCD TV though.

It also turns out that I can’t cut any straighter with a jigsaw than a hand saw. I think I need a lot more practise with all saws.