From: Main, Kerry [Kerry.Main@compaq.com] Sent: Monday, September 04, 2000 5:48 PM To: Info-VAX@Mvb.Saic.Com Subject: RE: Big AlphaServer Sale Andrew - >>> NT sales person says how can you buy OpenVMS it does not scale, OpenVMS sales person says oh yes it does, NT sales person says so show me some OpenVMS cluster TPC-C benchmark results or for that matter any standard benchmarks running on an OpenVMS cluster that demonstrate its scalability and while you are waiting for the OpenVMS sales person here is a little light reading prepared by Microsoft and Compaq.<<< The OpenVMS Sales person would provide the following Customer testimonials: http://www.iseoptions.com/about/technology_compaq.html (new OpenVMS Customer that just went live in May of this year) http://www.compaq.com/alphaserver/gs/quotes/etrade.html (recent GS Series upgrades) http://www.compaq.com/inform/issues/issue31/features-184-b1-3.html (Oracle and OpenVMS) http://www.e-dbms.com/analysts/2000/benchmark.html (Feb 2000 performance testimonial from Intersystems Cache database product on OpenVMS.) More testimonials on the new GS Series systems available at: http://www.openvms.digital.com/gsseries/quotes.html Now before you jump up and down shouting about where are the OpenVMS TPC-C numbers, it might be wise to note what the Sun press releases have to say about TPC-C.. :-) (a pretty good number by the way) quote extract "Recently, the TPC-C benchmark has fallen under sharp criticism for failing to keep pace with the times. Two separate cluster results reported by IBM and Compaq confirm that it can no longer be trusted as an accurate measuring for ad-hoc workloads. It's well-understood in the technical communities that TPC-C no longer represents current customer workloads since the transaction load that its models are made of are small, primitive and disconnected transactions. While this model was acceptable for the workloads of the late 1980s, it misses the mark for the object-based, integrated application environments of today's dot-com world that are part of the system design criteria at Sun. Customer workloads nowadays require a more ad hoc workload than the TPC-C specifies. This is why many customers compare systems using more complicated transactional workloads such as SAP-sales and distribution, SAP-banking, SAP-retail, Peoplesoft, or Amdocs billing. Areas in which Sun has excelled and maintains an industry-leading position. The simple nature of TPC-C transactions makes them well-suited to partitioning database tables across multiple systems as epitomized by cluster results on IBM's 32-node Netfinity System and others. This is possible because over 99 percent of TPC-C transactions are localized, while in the real world of customer environments almost no transactions are localized." end quote So, in light of the above, would your recommendation to the OpenVMS organization be to do a big TPC number or instead focus on Customer Application benchmarks (like the Interstate Cache database press release noted above) ?? :-) Regards, Kerry Main Senior Consultant, Compaq Canada Professional Services Voice : 613-592-4660 FAX : 819-772-7036 Email : kerry.main@compaq.com -----Original Message----- From: Andrew Harrison SUNUK Consultancy [mailto:andrew.nospam@uk.sun.com] Sent: Wednesday, August 30, 2000 1:53 PM To: Info-VAX@Mvb.Saic.Com Subject: Re: Big AlphaServer Sale Christopher Smith wrote: > On 30 Aug 2000, Jordan Henderson wrote: > > > Yes, you are confused. Compaq marketing also promotes OpenVMS clustering > > as the most advanced available. > > Where? > > > Nowhere in any of those references that you quote will you find that > > Compaq or Microsoft are particularly promoting the shared-nothing > > clustering of Windows 2000 as being highly advanced or the best > > way to cluster. > > In support of andrew, here -- by marketing the stuff at all, the public > being as ignorant as it is, compaq is effectively tricking all of the phbs > into believing that windows clustering *is* advanced. The only way they > could counter this would be to market both solutions side-by-side, touting > the VMS solution as more advanced. > That is the whole point. OpenVMS clustering is clearly more advanced than MS clustering and I for one would be much more confident about building a reliable scalable cluster using OpenVMS as the platform than NT/Win2000 but Compaq and Microsoft would have you believe from their marketing/benchmark reports that I would be making the wrong choice. Look at what MS say about the benchmark that Compaq paid for. http://www.microsoft.com/SQL/productinfo/tpc.htm They refer to it everywhere in their SQL-Server literature and it gives you the impression that you could build a large transaction system with it. But its a shared nothing system where almost all the "cluster" functionality is supplied buy SQL-Server. Who gave MS this ammunition none other than Compaq and they spend a load of money to do this. They had to run the benchmark twice because TPC were not happy with the origional run, they ran the benchmark on 4 different configurations each of which under TPC-C rules would have to be built specifically for that benchmark they used a shared nothing architecture requiring much greater setup/configuration costs etc etc etc. Let me quote you from MS's press release you will love it, BTW this is after the second run of the benchmarks after the first result had been rejected by TPC and Compaq had withdrawn it. "But like Rocky rising from the canvas to score a knockout, Microsoft and Compaq have released new performance results that not only address the TPC?s earlier concern but also exceed the earlier Microsoft Compaq performance record and provide the industry's best price/performance on clustered hardware." So Jordan please please don't bother trying to wriggle Compaq out of this. On one hand you have a couple of OpenVMS advocates who may or may not work for Compaq advocating an OS based approach to providing scalability and reliability in a clustered environment and on the other hand you have Compaq spending a huge chunk of money to help Microsoft market an applications led approach to solving the same issue. If Compaq were serious about proving OpenVMS's cluster scalability they would have invested in a similar benchmark for OpenVMS clusters, the fact is they never have making the comparison with what they have been prepared to do for win2000 even more damning. Put yourself in the position of a unknowing customer being sold an OpenVMS cluster or an NT cluster. NT sales person says how can you buy OpenVMS it does not scale, OpenVMS sales person says oh yes it does, NT sales person says so show me some OpenVMS cluster TPC-C benchmark results or for that matter any standard benchmarks running on an OpenVMS cluster that demonstrate its scalability and while you are waiting for the OpenVMS sales person here is a little light reading prepared by Microsoft and Compaq. Regards Andrew Harrison Enterprise IT Architect