https://slotsdad.com/ - casino online slots

Iz a Puzzlement…

by Administrator on April 25, 2014

I always loved the King and I musical, the one with Yul Brenner as the King of Siam.  I liked the scenes when he was cogitating on some imponderable, like why President Lincoln had no elephants in his army.

king_and_i_yul_brynner_small“Iz a puzzlement.”

I am feeling the same way as vendor after vendor reports flat line growth in their sales of external storage products (arrays, that is).  On 4/24, Storage Newsletter reported that EMC storage reventues were down 22% Q/Q, 3% Y/Y.  This followed a 4/17 report from the same source of a “catastrophic” drop in IBM storage hardware revenues, off 23% Q/Q.

Why EMC’s numbers are not catastrophic at 22% decline, while IBM’s are considered such (at 23%) is a point of some puzzlement for me.  IBM is a much more diversified company than EMC, so I would think that a drop in storage sales here is likely made up for by a hike in Big Data analytics software or mainframes or whatever…  But Mr. Maleval calls them like he sees them.

Anyway, I want to match up these reports and others like El Reg’s reportage on sagging storage sales in networked storage and the story about Wikibon (carried in April by the Register) pronouncing the death of enterprise storage in favor of little server-side cobbles like VSAN.  I am left with a big question.

What are companies doing about the explosive data growth that IDC and Gartner were pegging just last year at 300% to 650% per year in highly server virtualized environments.  How/where are we storing all of those bits?

Iz a puzzlement.

Data doesn’t store itself.  We don’t manage either data or storage very well, so we definitely aren’t making the best use we can out of the assets we have.  Last year, IBM’s Storage Czar said on stage that everyone deploys too much Tier One storage and that had to change before it bankrupts the IT budgets of companies.  Last year, storage clouds claimed a huge uptake in storage capacity, but closer inspection revealed that this capacity was mostly being consumed by compute cloud operators, not by “users.”

If I read the back and forth correctly between Chuck Hollis and George Crump (see previous post), the VSAN thing is representative of the problem created by VMware in terms of capacity demand growth.  While VSAN can be deployed as a mirrored pair of storage servers, Howard Marks and others correctly point out that an HA configuration really needs three VSAN toys.  So, that accounts for IDC’s 300% capacity growth curve from last Summer.  Gartner said server virtualization was jacking up capacity demand by 650%, presumably calculating in not only replicated data but also the multiple backup copies and the additional replication that is made behind any server that might possibly be tasked to host a give workload.

Why is VSAN necessary — other than to solve a stupid VMware-caused problem, that is?  Iz a puzzlement.

Why not just virtualize your existing storage (no, not “software-define” it — which some purists insist does not include the aggregation of capacity or its presentation as virtual volumes, but actually virtualize it) and associate virtual volumes with your VMs.  That way, when the VM moves from server A to server B, you don’t need to fix the application’s back end connections to its storage from each new perch.  The virtualized storage provider will do that for you.  That’s how I do it with DataCore Software and it is presumably how I would do it with an IBM SVC.

Iz a puzzlement.

I would love for real world operators to tell me how they are provisioning storage to the increasing storage burden if they aren’t buying new capacity.  Are we throwing our mission critical data onto big SATA hard disks inside our servers or in DAS chassis?  How do you handle the vendor lock-ins that constrain your ability to replicate data from one brand name box to another?  What kind of bandwidth consumption are you seeing in lots of DAS to DAS replication processes?  How do you test intelligently whether you have all of the data you need to failover or to host a VM on a specific physical server?

See how the questions build up when you begin with the premise that data is growing exponentially, but storage sales are flat?

Iz a puzzlement.

I’d really like to get IBM’s take on this when I am at IBM Edge 2014 from May19 through May 23.  Hope to see some of you there.

Edge 2014

 

Note:  The questions above are in no way vetted, approved or otherwise monitored by Big Blue.  However, I will be supporting their social media efforts and delivering five training sessions at Edge, activities for which I am being compensated.  This disclaimer is required by the FTC.

 

Previous post:

Next post: