Simply changed

It’s almost two weeks ago since our entire data base replication laid down its service. The master suddenly started to jerk again two days ago. It simply forgot data bases although we have practically changed the whole servers’ hardware on 17th June.

I had an unpleasant feeling anyway because so appropriate hardware problems like a crashed CPU or RAM bar with defective sectores aren’t eventually available. If the mainboard doesn’t just like the network interface card, our hands will be tied. Driving to the computer centre every week to stroke the boxes- which temporary helps at least first-hand and also according to statements of one engineer in the computer centre- is no real solution for us.

We contacted again the kind ladies of the section “Distribution corporate client” already on 17th June to inform about current servers as substitute for the old data base servers. An offer- which we couldn’t quasi disapprove- was available on 24th June after some emails and telephone calls.

Our complete new data base cluster with recent hardware has run for two hours. An Intel Pentium Core2Quad Q6600 with 4 cores each 2,400 MHz and 4 GB RAM runs as Master. An Intel Pentium Core2Quad Q9300 with 4 cores each 2,500 MHz and 4 GB RAM even works in the slave. The bottlenecks were always during the calculation so far. That’s why we decided on thicker processors for both machines.

The announced optimization works have already produced first results. Meanwhile, our servers have more write than read accesses. That’s insofar good because our software is already able to satisfy more than 90% of all reading information retrievals out of the memcache memory. We can trap several information retrievals- which have directly gone to the data base until now- but there’s still no demand.

And now we’ll relax and curiously trace the computer monitor(ing) how the both new ones work on their first day in the toFOUR family.

Sunny regards from Thuringia,
Markus & Tim

is powered by toFOUR GmbH