We are testing UD3.0.0.2 TCP/IP on novell 5.1 and getting significantly poorer performance than with the existing IPX driver. Initial connection seems to be about twice as fast. Iterative access (reports, non-indexed lookups, etc) seems to be 2 to 10 times slower.

Any suggestions? Any benchmarks?

We are also converting a to the UD on a Novell 5.x network (5.5 for us). It is an Arev3.12 application.

If it start running 2 to 10 times slower, we will all be dead as the angry villagers will chase us up into the windmill and light it on fire.

I have heard that IPX is slower on larger networks and faster on smaller networks than IP. Is there a network size where the new UD with IP is faster than the LH with IPX?

I mis-stated, our Novell is v5.1 also.

Anyone? RTI? Is this something we should be checking into? Where can I find the benchmarks referenced in the UD posting:

"Increased speed performance.

The Revelation Network Products are designed to enable database activity to be off-loaded onto the file server. This architecture dramatically reduces the amount of network requests. The reduction of network requests translates to increased performance. Benchmarks show a dramatic performance increase on local area networks (LANs), and an even greater increase on wide area networks(WANs)."

Ralph,

It is pretty much common knowledge that the Network Products speed up the use of a Revelation application, and a few generations of benchmarks have proven that true.

As for benchmarks for the Universal Driver, we are working on them. To run the benchmarks for a new product that covers three Network Operating Systems is a pretty substantial undertaking. For each permutation (i.e. OI 4.1.2 on a Windows XP workstation on an NT 4.0 Server), we need to run 84 separate benchmark tests. Multiply that by three versions of OI, three WS operating systems (at least), and (with the different versions of the UD on Windows) approximately 14 Servers and you are talking about a lot of testing!

I spoke to Craig and mentioned that in some of the preliminary tests running the UD on Netware, it looked to me that the UD was slightly slower than the NLM. That was based on no in-depth benchmarking, just a quick look-see. I have heard that at some Netware UD sites, the UD was slowing at peak work times. On the other hand, I have also heard from the field that people were seeing an amazing speed increase using the UD on a Netware server - in some cases reports were running 7x faster than with the NLM - and this was across multiple sites.

The benchmarks should be the proof in the pudding - just as they have been. We are close to starting the benchmarks on the Novell Servers - and they should go fairly quick.

In the meantime, we are working with Craig to see if we can pinpoint where his problem is.

Hope it helps,

Kevin Ruane

The presence of multiple protocol stacks will slow things down. So TCP/IP and IPX (even in the incorrect load sequence on the workstation) will adversely affect performance.

I don't know if this will apply but it worked for one of our

clients that experienced EXTREME slow downs after installing the

Universal Driver. They are using Novell 6.1sp3.

From each workstation, they right-clicked on the icon that called

up our AP with OINSIGHT.EXE and clicked "Verify". Apparently, this is

part of the Novell Application Launcher. After that, the program

"screamed" it was so fast. This option is not normally available, only

with those using the Novell Application Launcher.

Hope this helps.

Brock Prusha

Sr. Systems Analyst/Project Leader

[email protected]

Mapcon Technologies, Inc.

View this thread on the Works forum...