I was told repeatedly by several different IBMers that R5 Domino on the mainframe scaled so much better than Domino on the Windows platform. We had a lot of performance issues when we started on the mainframe, moving to z/OS helped, but as long as I am reading this correctly, it looks as though the chart shows R5 scaling better on Windows that on the mainframe. When you hear it from people that are suppose to be experts on the subject matter you tend to believe it, but then you see otherwise in print and wonder what to believe.
Excerpts from a Redpaper (3634) "Upgrading to Domino 6 Performance Benefits"
Figure 1-1 shows a stacked column chart that combines the number of supported users (bottom portion) with the percentage of remaining available CPU (top portion). Taken together, they indicate how hard a particular system had to work to achieve the specified number of users. Each set of data has been
converted to use a scale of 1 to 100. So, the highest number of users possible on the chart is 10,000 (100 scale times 100 users in a set). The highest remaining CPU available would be 100%. Therefore, a system supporting 10,000 users and requiring no CPU to do so (impossible, but it would be nice) would reach to the top of the chart at 200. So, in any case, the higher the column the better the result.
Looking across the chart you can see that Domino 6 was able to reach the goal of 10,000 users (each bottom portion of the chart for Domino 6 reaches the 100 scale line, meaning 10,000 users) on each platform while maintaining a 1 second or less response time. In each case, Domino 6 also had a good deal of CPU resource left over (in each case 50% or more looking at the size of the top portion of the chart). Simply put, Domino 6 performed better than R5 on all four tested platforms.
The z/OS platform added 3,000 users with Domino 6 and still used less CPU than R5.0.10. And even when the number of users did not increase greatly,
Domino 6 utilized much less CPU. The Solaris R5 system reached 9,000 users but had very little CPU processing power remaining. The same Solaris hardware running Domino 6 reached 10,000 users and had 30% more CPU processing power left than the R5 system.
On the OS/400 platform, a single Domino 6 partition running on iSeries hardware had 12% to 26% more CPU processing power left than the R5.0.11 partition supporting the same number of users (3,000 in one test and 8,000 in another). That translates to a 30% improvement in CPU usage. Domino 6 also had better
response times than R5.0.11 during both test runs.
At a glance, it is clear that just upgrading your servers from R5 to Domino 6 will allow you to support a larger number of Notes client users. Alternately, by moving to Domino 6 and keeping the user load the same, you can reduce the CPU workload burden on a server operating near peak CPU utilization.
What about upgrading Notes clients? The Domino 6 server supports mixed clients, so after completing an upgrade of your servers, you can then upgrade your Notes clients. Once you upgrade to Lotus Notes 6 you will be able to use a powerful new feature, Lotus Notes Smart Upgrade (see the IBM Redbook Upgrading to Lotus Notes and Domino 6, SG24-6889 for details). Domino 6 gives administrators the ability to post new client builds and have employees automatically prompted to install them with the click of a button when they connect to their home server. Your employees can enjoy the full functionality of the Notes client while you benefit from a centrally managed upgrade process.
Since future upgrades happen automatically, you can immediately exploit the benefits and performance improvements of each new release of the Lotus Notes client.
Consistently better performance results across all platforms, coupled with the ease of upgrading the Lotus Notes 6 client, means an IT department can upgrade from R5 to Domino 6 and not only expect quantitative performance improvements from their Domino servers, but also benefits such as productivity efficiencies.
No comments:
Post a Comment