IBM's CFO Mark Loughridge put on his product marketing cap during an earnings conference call today, revealing the imminent arrival of new Unix boxes and mainframes. According to Captain Finance, IBM will push Power6 chips into low-end Unix servers by the end of this quarter. Customers should welcome this charge, since IBM has …
But will they...
run Vista? I don't think so. So buying one is just a political statement.
@ Anonymous coward
Why on earth would you want to run Windows Vista (or any other version of Windows) on a Unix server? These machines are designed to run Linux and AIX, not Windows. The Power6 chips excel at parallelisation and massive throughput, and are designed with SMP systems (those with many processors) in mind. In other words, these are great machines for computationally intensive tasks that can be split into many smaller tasks that run simultaneously, and are also excellent database servers.
I doubt - since running an ERP for a 1000+ users (typical U*X workload) consumes less resources than a single-user Vista installation.
But will they...
have motion sensitive controllers? I don't think so. So the Wii is still better.
So that "virtualisation... technology for moving live applications across boxes" will finally give IBM something close to the automated capability that hp-ux has had with MC/ServiceGuard, Process Resource Manager and Global Workload Manager for years, then?
And I think they'll find the IBM mainframe sales dropped 15% because the customers were switching to cheaper systems running UNIX or Linux....
I'd like to know...
...how they stack up to what Apple will be coming out with this year...we never seem to get these comparisons from the industry press or any (non Apple) vendors. What are they afraid of?
MC/Service Guard vs Virtualisation
MC/ServiceGuard does not move "LIVE" applications. It just can be compared to HACMP (clustering software from IBM).
I suppose it depends on your definition of "live", but there is no way I would compare HACMP with MC/SG, it's like comparing the Trabant with a Mondeo. Have IBM actually got HACMP supported with AIX6 yet, which is the version you need to take advantage of Power6? If I remember, it wasn't ready for the Power6 launch which is one of the reasons IBM had to keep on pushing Power5+ across most of the range. Of course, with Integrity I can do my virtualisation and clustering with hp-ux, Virtual Server Environment, and MC/ServiceGuard, but then I can also host Windows or Linux in the Virtual Machines on the same box. Or, I could just use hardware partitions for hp-ux, Windows, Linux and OpenVMS as fully electrically isolated instances within the same chassis, which IBM can't do to properly with any version of Power with even just AIX.
Please search and read a little before posting.
Google may be your friend, search for "HACMP AIX 6.1" and you will get :
let's try another search : "HACMP 5.4.1" and you will get :
summary : HACMP supports AIX 6.1 and WPAR
2) Live Partition Mobility
with google search "SAP live partition mobility" and you will get
a true and real moving partition from one physical server to another one without (I said without) any shutdown of application nor AIX partition. That's whay the word used is "LIVE"
another search , may get you the following web page :
hope this help !
HACMP and AIX6.1 - I see this is only dated November 2007, so you're only six years behind hp-ux in providing a vpars copy. I see that the HACMP doc talks about "failover", so just like every other cluster out there then and only how many years behind? Did I mention that HP have had ServiceGuard support ready for each release of CPU and each release of hp-ux since..... well, longer than I can remember, actually!
Nice LPM demo vid, but if you watch then you will see the SAP application is temporarily suspended during the process, so I think I can safely file your "I said without any shutdown" under the same IBM guana-guana bin as "seamless integration". The demo is dated November and "proof of tech" so I assume the IBM marketeers have plenty of time left to work on the salespitch. Which explains why our pet IBM salesguy gave his pitch without (I said without) mentioning this in December. But then he knows I have a nasty habit of making him prove what he says, which has cut down enormously on his little chats!
In the hp-ux world you would just use Global Workload Manager to shift the SAP processes around to balance out load throughout the cluster without having to shutdown or migrate whole instances. Or I could just move resources around between virtual instances to give the "hot" instance more grunt. In fact, this seems a very complex way of achieving what most OSs manage through ordinary live-live clustering. Oh, hold on a sec - I bet this needs only five months of IBM Global Services to get running, not six like the old! ;)
After doing some more reading and playing with my new copy of RHEL5, I have come to the sneaky suspicion that the LPM capability is just a rehash of Xen's virtual machine migration!
/me drops email to Eben Moglen.....