Following on from our Sage 50 and Vista testing we have now had the time to carry out Windows 7 performance tests.
The good news is it doesn't seem to have the performance sapping cache effect of Vista and performance is much the same as XP (even though it's running on an iCore-5 as against XP on a Pentium P4 2.8ghz. Oh well).
Interestingly, swapping from 100mbit Ethernet to 1gb made almost no difference to our Sage performance test results although a straight 25mb file copy was around 4 times faster.
Monday, March 14, 2011
Monday, March 07, 2011
Sage 50 and Vista - slower than a slow thing
Ok, I know Vista is now water under the bridge but there are still a lot of these misbegotton systems about.
Recently I have been investigating the use of NAS devices with Sage Line 50 and spending a lot ot time analysing the performance using our Sage testing tool, Transfix.
What was strange was that the first run of the test was quite fast, but each subsequent run became slower and slower until it botommed out at about 3 times slower than the initial test.
Odd.
To cut a long story short the problem is with Vista caching the data to RAM.
On the first run, all the data (in this case a 30mb Sage file) is read across the network. Each subsequent run loads more of this data in the Cache until eventually there is no network traffic at all when the file is read ; it all comes from the cache. You can observe this by monitoring the NIC Status and using Task Manager to see the cache grow.
Sadly the Sage SDO is around 2.5 times (60 seconds vs 150 seconds) slower reading directly from RAM than from the network.
The file is only cleared from cache when a process causes the the file to be updated which then restores normal performance until the file becomes cached again. There does not appear to be any way of controlling this behaviour.
I have no idea if this affects other programs but certainly Sage doesn't like it.
XP doesn't have this problem and I haven't got around to testing Windows 7 yet.
Does nobody test this stuff?
Recently I have been investigating the use of NAS devices with Sage Line 50 and spending a lot ot time analysing the performance using our Sage testing tool, Transfix.
What was strange was that the first run of the test was quite fast, but each subsequent run became slower and slower until it botommed out at about 3 times slower than the initial test.
Odd.
To cut a long story short the problem is with Vista caching the data to RAM.
On the first run, all the data (in this case a 30mb Sage file) is read across the network. Each subsequent run loads more of this data in the Cache until eventually there is no network traffic at all when the file is read ; it all comes from the cache. You can observe this by monitoring the NIC Status and using Task Manager to see the cache grow.
Sadly the Sage SDO is around 2.5 times (60 seconds vs 150 seconds) slower reading directly from RAM than from the network.
The file is only cleared from cache when a process causes the the file to be updated which then restores normal performance until the file becomes cached again. There does not appear to be any way of controlling this behaviour.
I have no idea if this affects other programs but certainly Sage doesn't like it.
XP doesn't have this problem and I haven't got around to testing Windows 7 yet.
Does nobody test this stuff?
Subscribe to:
Posts (Atom)
