RE: Looks like the challange has been made
<Yawn> And still waiting..... Actually, getting bored of waiting, so I suppose I'll just have to poke fun at Aelfric. By the way, did you know there was supposedly a King Aelfric of Kent? Well, according to St Bede, but modern historians have described him as a fantasy king created to fill a gap in monastic records. How appropriate!
"....Matt Bryant will have to answer yes or no to the "FUD" on Tuesday...." Considering you seem very sure of yourself I'm going to have to assume you are actually in breach of an hp NDA agreement. If not, then it must be just more fantasy. If it is in breach f an NDA then I'm betting you have tried to paint as bad a picture as possible, so let's look at what you have to offer....
".....1) HP will only announce blade Tukwila systems...no rack based systems...." I've been told it is an Integrity BLADES event, so I would only expect blades to be announced. Do IBM announce their mairframes at Power blade events?
".....2) HP will not mention Linux in regards to Tukwila systems..." Lol! Even if you are leaking real technical news, you can only guess at what the speakers will actually say, or are you now claiming to be their speechwriter?
".....3) The Tukwila blades will not have SX3000 glue chips so the 8 socket system will require chip hopping as Tukwila only has 5 QPI's...." Why would you expect a dual-socket blade to use a four-socket chipset? If BL860 is to have 8 cores as you say then it will be dual-socket, and then if the BL870 is to be two BL860 s then that's two dual-socket motherboards, not a single four-socket. Do you even think before you trype (sic)?
"....4) vPar will not be supported wtih Tukwila...." They're announcing blades at this event. The users I speak with are using IVM with the current Inetgrity blades rather than vpars, so why you think it is news the new ones won't is beyond me. Oh, I get it - you couldn't think of a real issue so you decided to make one up! I suppose you don't want me to predict that Inetgrity Virtual Machines (whcih offers CPU or sub-CPU granularity of virtulaisation, so better than just vpars) will be supported out-of-the-box?
"....5) The "Scalable bladelink" is a big connection on the front of the blades which will restrict airflow..." Assuming you mean the link between blades acting in a single OS image, that would only require a narrow ribbon cable, not a great big item, so very little impact on airflow. How do I know? Well, if you look in the back of an SD64 you can see the inter-cab link that carries traffic for SIXTEEN cellboards (each equivalent to a BL860, so a lot more traffic than between just TWO blades), and that is just a brown ribbon cable about 5-odd cm wide. Then again, I suspect you have never even seen an SD64, let alone looked inside.
"....6) The BL860 is 8 core, 24 DIMM slots 2 hot swap SAS HDDs 4 GBe and 3 PCIi...." Hold on a sec, you're supposed to be making the hp kit look bad! Didn't you realise that's far better than the PS701 Power blade's spec? After all, the PS701 only has sixteen RAM slots, 2 mezz card slots (PCIi), and no hot-pluggable disks, and only two LAN links! Lol, I can see why EDS "let you go", you just keep doing the opposition favours!
"....7) The blades will only supports 4GB dimms at release, 8 GB is 3Q 2010, 16GB is 4Q 2010..." The IBM blades do have 8GB DDR3 DIMMs (well, when you actually get them, which will probably not be for a while yet). Of course, you have to drop the bus speed from 1GHz to 800MHz to use the 8GB DIMMs in the Power blades - I wonder what impact that has on performance? Then again, seeing as the Power blades will have 3GHz crippled cores (nowhere near as fast as the Power rack servers will eventually get), I don't suppose it will make that much difference having everything in the blade crippled. And the BL860 will have more RAM slots anyway.
"....9) HP will try to slow the defections off of HP-UX by offering socket based pricing but since Tukwila is only 4 cores per chips its not that big of a deal....." A bizarre statement. You can already get the base hp-ux 11i v3 OE for free with any Integrity server. The other OEs (Operating Environments - think software bundles with a single licence) have always been priced per core, so surely you have just told all the hp-ux base they can expect a 50% cut in OS costs if they upgrade? Another favour for the enemy - you sure you really want to try this FUDing thing, you just don't seem very good at it!
".....10) Open VMS is not supported till Q3 2010...." <Shrug> Not an OpenVMS bunny so doesn't really affect or bother me. But, seeing as IBM failed to poach even a fraction of the Alpha-VMS base when Compaq was bought by hp I can't see there suddenly being a rush to Power now.
".....11) nPars is 2Q 2011 and vPar is NEVER...." Ah, I see that you don't think before you trype. For those that may not know, npars are hardware partitions, i.e. the means to link more than one Integrity cellboard together. But, Aelfric already said "...."Scalable bladelink" is a big connection on the front of the blades....", which kind of implies the ability to hardware link blades (i.e., create npars) is already there at launch. As for vpars, see my comment above about IVM.
"......12) if you want to upgrade a 860 to a 870 or a 870 to a 890 you have to wait for the field upgrade kits in Q1 2011...." Or you could just order an 870 instead of an 860 in the first place. It's not unusual for upgrade kits to arrive well after a launch as most customers don't think about upgrading for at least a year AFTER they buy kit. We certainly don't buy servers and immediately upgrade them. Then again, we do proper planning an POCs, whereas IBM seem to expect customers to just take Power on the basis of some SPEC benchmarks and then have IBM Global Scewups onsite for a year trying to get the solution working. I suppose IBM have to supply upgrades immediately given their awful scoping work.
"....13) Want to do FCoE? You have to wait till Q2 2011....." Could that be because the FCoE standard is not expect ed until April 2011 EARLIEST? Any FCoE kit you buy now could be rendered obsolete overnight if the standard differs to what the vendors are guessing it will look like. And that also means any FCoE kit bought now will also likely not work with another vendor's as they are all doing their own thing. In the meantime, hp will sell you an FCIP solution using MX400 routers if you really don't want darkfibre links, and those FCIP routers work fine with the existing hp blade chassis switches. So, another non-point.
No need to wait for Tuesday. The answers to the above show that EDS were only too smart to treat you like the trash and bin you. Now, can the IBMers please put up a real troll? Maybe three or more of you can get together and pool resources, make a real effort?
/SP&L