RE: Anonymous Sunshiner / Marketeering Droid / ignoramus
Where to start, just so much male bovine manure to cover!
"Hey Matt, who cares who had what first?..." The market cares. You see, they like tried and tested solutions, and HP, IBM and Dell have built up customer belief in their x86 products over time, which is why they have been successful when moving into blades as they could play on that trust already there from their x86 racked servers. Sun has never had that, and now they have to forge new relationships with people outside the traditional SPARC arena. From the market data it looks like that isn't going well and the Sun x86 and Niagara are just cannibalising existing Sun bizz rather than gaining market share. This does not bode well for Sun, especially as the resellers they will look to sell the x86 servers (and hence all those blade chassis the Niagara blades will need to plug into) will be experienced IBM, HP and Dell resellers, all unlikely to go to the trouble of switching accounts from HP, IBM and Dell x86 products.
"....The real story is the T2+ blade which HP nor IBM have anything even close...." This is true. All the HP and IBM blades, whether Xeon-, Opteron-, Power- or Itanium-based can actually run heavy threads and therefore offer the ability to consolidate real enterprise apps such as Oracle and SAP off old RISC servers into shared blade chassis farms. HP, IBM and Dell wouldn't think to produce a limited non-product like the T2+ blades, so you are quite right there. Niagara can only do consolidation for the thinnest of old SPARC apps, but is very good for light webserving. It's just such a shame that webserving is a market already dominated by Lintel/Wintel, from HP, IBM and Dell, where those vendors have the established relationships and reseller backing, and Sun has SFA....
"....Look at the new benchmarks being won with the T2+. It's pretty impressive. Especially versus HP. HP is quickly falling off the performance curve...." Really? So where are the SAP, Siebel, Oracle or MS SQL figures for T2+ and how do they compare to a Superdome? Oh, I forgot, T2+ is a memory-crippled T2 which barely scales to two sockets, and even then can't handle real enterprise apps. Maybe we should compare it to a Xeon ProLiant? Then again, maybe not, seeing as I can't find any enterprise application benchmarks for T2+ using respected test tools such as TPC-C or TPC-H which customers use to compare systems. SPEC has an even funnier story - Sun has submitted test results for the rate versions which test the whole system, where they can make the most of the high core count, but have zero results entered for the individual thread tests (gee, I wonder why....).
"....In HPUX a badly behaving application can take down the other applications. With Solaris Containers the bad application is contained within its own container, causing no issues to the other containers..." You obviously have zero knowledge of hp-ux's partitioning technology. Even going back to the old PA-RISC servers I had software partitioning, where each partition had a completely separate OS image, though hardware could be shared, which is better than containers can offer now. And I also had hardware partitioning on the cell-based PA-RISC servers, which meant I had a share-nothing solution with real isolation. With containers, if the underlying OS goes down then all the containers go down, period. Add on top of this that hp-ux had resource-based partitioning (PRM) and SLA-based management (WLM) tools in the OS which could stop a misbehaving app hoggign resources even before Slowaris got containers and you will plainly see HP where years ahead of Sun on UNIX virtualisation. Oh, sorry, I mean you will see after you loosen up the Sunshiner Blindfold (TM) which is probably also restricting the flow of blood to your brain.
"....Fujitsu & Sun coming up with a Sparc based processor thats now doing Instruction execution retry, a feature not seen outside of the mainframe in reality. Hang on, lets think about your dearly beloved HP products again, ah, yes, Itanium is not quite up to Instruction retry yet is it, hmm...." Latest Sun feature sell? Simple way to beat it, just like all the previous Sun feature sells - go to a shoot-out, run real data to mimic the real environment, including deliberate errors, and see how the respective systems stand up. So far, I've happilly defeated all Sunshiner attempts to uproot our Superdomes by this simple method, and usually because Sun drag their feet over going to a shoout-out they know they'll lose. My most entertaining bit of Sun humiliation was running an Oracle-linked billing app faster on hp-ux on Itanium than Slowaris 10 on SPARC64, then booting the partition into Red Hat Linux with Transitive QuickTransit and running the old Slowaris 9 binaries faster than the SPARC solution on Red Hat on Itanium! I will admit we did later look at replacing the Xeon blade web front-end to the app with T2 kit, but in the end we decided the added cost, separate support contract and admin burden didn't warrant the minor gain in performance, even before we looked at the shakey roadmap for Sun.
"....Erm, Sun, first company to tread the "many cores" instead of "mhz" route...." Ignoring AMD's Opteron? And you went dual-core with Ultra-SPANKED, and your business died because single-core CPUs in HP and IBM servers outperformed it at a lower price and with far better RAS features. Like many Sunshiners, you have still failed to grasp the fact that the company that makes what is closest to what the market requires will win. Making "wowie-zowie" but low-performance, expensive products with limited market appeal has become a Sun speciality. Even Dell are better at the market prediction bit than Sun, whom have happilly wasted time on numerous dead-end projects before limping after the other vendors which have left Sun far behind. Suns storage portfolio is late and thin compared to HP and IBM, their blades are even later and just as limited, and their "Java" thin-client answer to Windows on the desktop is so low in penetration it lags Novell and Apple in marketshare! Sun is the least diversified of the commercial UNIX vendors, which is why the decline in the UNIX market has hurt it worst. You act all snobby about HP printers, but the truth is HP will be making a profit off printers long after Sun has been broken up and sold off.
Biting the hand that feeds IT
