About that IBM hardware revenues dive: Blame storage, says CFO IBM's storage hardware revenues declined by 19 per cent year-on-year in its third 2015 quarter, continuing a four-and-a-half year fall. IBM_Storage_HW_revs_to_Q3fy2015 IBM Storage hardware revenues to Q3 2015. IBM CFO Martin Schroeter put IBM's revenue declines in the context of IBM switching from declining businesses to …
No!
The future of IBM storage is Software-defined
IBM Spectrum Accelerate; a VMware deployable image that turns internal HDDs/JBODs and Flash Cards into a storage cluster where capacity is made available to any VM host via vSphere or OpenStack (Cinder) over iSCSI.
IBM Spectrum Scale; an installable driver (not VM image) transforming local HDDs, Flash PCI cards and RAM cache into a distributed storage continuum with global consistency, able to sustain data rates of 2k files/sec or 30Gb/sec and above. In contrast to hyper-converged appliances shows that only 3-5% of clustered servers should have local disks, yielding the optimum balance of access density versus metadata traffic.
IBM Elastic Storage Server; (Spectrum Scale-in-a-box); high-speed n+2 or n+3 erasure code data protection for JBOD, defined by policies and based on POWER8 architecture running Linux.
Obviously IBM will maintain DS8K for the mainframe, as it continues to develop and enhance it's mid-range Storwize family that formulate not only it's own converged system offering in PurePower, but also the converged system offerings of it's partner's Cisco (VersaStack) and Lenovo (Flex Systems), whilst exploiting the growing flash storage market they bought into acquiring TMS to create FlashSystems.
They did not exit the x86 server business only to re-enter it because of flash-in-the-pan technology.
Wow - that reads like buzzword/bullshit bingo!
Also, IBM are choking in the storage arena because products like the "mid-range Storwize" suck a fucking incredible amount of rectum. I mean they're just abysmal! Put me in front of a v3700 or an HP MSA P2000G3 (there's an old beast), I'll go with the MSA every time.
Anon because we have just deployed a bunch of v3700s and I despise them and everything they stand for.
"Yikes, I guess the budget couldn't stretch to a FlashSystem V9000 then"
Nope. And neither can most people's. But having its arse handed to it by devices I was using 4 years ago is embarrassing.
Also, no SNMP? So I have to have somebody manually check the available space on a regular basis.
And the v3700 10Gb links don't work with IBM's own (re-badged) 10Gb iSCSI HBAs. Gotta use software iSCSI. Even after paying for all the hardware IBM say no.
I am actively campaigning against deploying anything from IBM or Lenovo now, except for T and X series Think pads.
Storwize and 10gb is a bad idea. The performance is undocumented, put it in FC you will like it runs very good. Our iSCSi performance sucks. Your rep is an a-hole for selling you iSCSI and not telling you the truth. If you don't have too many hosts you could even go SAS you can hookup up to 7 hosts it will still be much faster. Don't let the 10 in the 10gb fool you it's our iSCSI software stack that's the problem.
> The future of IBM storage is Software-defined
What do you mean? What is Software-defined?
Spectrum Accelerate: burns 6 vCPUs per host and needs minimum 3 hosts, For that gain no efficiency, no compression or dedup according to the Redbook. Bit of a joke product really, a copy of FreeBSD with ZFS would be a better proposition for most customers of this junk. (https://itunes.apple.com/us/book/ibm-spectrum-accelerate-deployment/id1006021968?mt=11)
Spectrum Scale: GPFS. Again. Has limited commercial success outside of a few science projects because it's brittle and needs scientists to keep it working. What is a continuum? Redbook is peppered with gems like: “Figure 1-1 shows a flow chart describing which IBM Spectrum Scale editions license to use depending on which purpose.” A flow chart for licensing! Nice!
IBM Elastic Storage Server: Prepackaged GPFS, so for those few who want GPFS it's wonderful. For the rest. would Sir like to look at VBlock / FlexPod for more sellable technology?
Spectrum +2: Tape drive occasionally fails. Bit like the IBM strategy.
> They did not exit the x86 server business only to re-enter it because of flash-in-the-pan technology.
Completely agree. They'll re-enter it when storage sales drop to near 0 and a new strategy is needed.
You want one management interface but then are considering GPFS?
You need to be a host-expert for each of the host types that are attaching, very familiar with scripting, and for backup you need to know other products like TSM. And good luck if you mix Windows in there as well.
Very 'business' and not at all 'techie'.
Maybe you meant some other type of 'single' interface?
This post has been deleted by its author
It gets worse, so you bought into to the whole VBlock/FlexPod thing to run Exchange & VDI when it could've been most cost-effectively in one of those big Data Centre's Microsoft has built or by bundling a couple of those Nutanix or Simplivity boxes together!
Who signs this stuff off in your place and does the Cisco rep get milk and cookies on each visit?
First of all the name GPFS ceased as of version 4.1 and became Spectrum Scale part of the Spectrum storage family of software products unified by a single management GUI, which all products in the Spectrum family have inherited from XIV, TSM is now Spectrum Protect. Spectrum Scale provides you with a Single Global namespace, anyone who has ever worked on a VB-Lock or FlexPod will tell you they are converged in name only.
They had better, because the current strategy clearly isn't working.
The so-called "transformation" is IBM's problem. They're getting into markets which already have leaders. IBM is picking up the scraps and as usual is trying to buy its way in through acquisitions. Of course, based on past experiences, they'll keep the IP, ruin the products, and all of the talent will leave.
Meanwhile parts of the business which are declining, at least when you're talking about revenues (storage isn't declining, not in the slightest - the amount you can get away with charging for overcomplicated licences is). So what is IBM's answer? Not streamlining those products, making them more efficient and getting the most out of their employees. Nope, the answer if you're not selling enough is to sack the sales force. The answer if your products are not keeping up is to slash development budgets. The answer if your product set is an non-integrated mess is to re-brand everything so that even those who sell them can't remember what the new name is, while throwing away the existing brand recognition built up over decades. Oh yes, and alienating all of your customers who know that you're getting out of the business and won't support them any more.
Your existing customers are also transforming their businesses. Why the hell do you think they will do so with you?
Having said that, it's not too late to change, but that time will come.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
