Timing Is Everything
by Hu Yoshida on Apr 23, 2009
In view of EMC’s recent earnings report, Barry Burke, an EMC blogger also known as “the storage anarchist”, posted an April Fools parody to deflect attention away from their earnings results and poke fun at an HDS promotion that any savvy storage customer would value.
On the other hand, the timing of the HDS promotion to provide free software to attach third party storage systems to the USP V storage virtualization platform was right on the money. While the promotion is not targeted at the EMC install base, it most certainly bodes well for timing since EMC abandoned their DMX users with the announcement of the Symmetrix V-Max — and someone had to rescue them (which is what HDS is happy to do). While V-Max added features like dynamic migration, it did not add storage virtualization so that these features could not be extended to existing DMX and Clariion arrays. They spent a great deal of time saying how the V-Max was x times better than the DMX, and by the way, the V-Max was for virtual server environments and the DMX was not.
These DMX users as well as Clariion users can now attach to the USP V, free of license charges, and enjoy the benefits of dynamic migration, dynamic provisioning (thin provisioning), and in-system replication. They can get rid of their EMC software and free up stranded or allocated-but-unused capacity for productive use. They might free up enough capacity to eliminate the need to buy any more capacity the rest of this year. But if they still need more capacity, they can add it through the end of this year under the same license agreement.
The timing is right to switch it on.
Comments (9 )
Now see, there you go again, Hu – spreading abjectly false information.
EMC in fact announced last week that DMX3 and DMX4 customers would be getting additional functionality this year as well, and the product they bought before the world knew about the V-Max is no less valuable today than it was the week before.
And you’re throwing stones while giving away free software in a feeble attempt to move the aging trapped-in-a-backplane USP-V after failing to deliver the promised USP-V clustering for the third year running.
Pot. Kettle. Black.
So tell us, just how are those UVM customers supposed to non-disruptively upgrade to the next-gen USP-V2 (you do have a next-gen coming, right?)?
Given no credible answer to that question, nobody in their right mind has chosen to accept the response time penalty, connectivity & maintenance costs and management complexity of adding ANOTHER layer to their storage infrastructure – a layer they have NO HOPE of transitioning out of without suffering the pains of Yet Another Migration.
Fact is, DMX4 customers are well served by the DMX4, even as they look to the V-Max for their next purchases to take them places that Hitachi hasn’t even imagined yet.
And be sure, most of the world have noticed that Hitachi was nowhere to be found in this weeks’ vSphere announcements. It’s going to take a lot more than a SRM adapter to be a player in the virtual data center – how can you claim you’re taking customers into the future if you don’t even show up in the present?
Bottom line is this – today’s parody was meant in to be in respectful fun, but judging from the unprecedented response from employees, customers and industry analysts, I suspect I may have hit a little too close to home.
If so, my apologies. I’m sure the economy hasn’t singled out EMC (or IBM, or Cisco, or Intel, or …). Hopefully we all make it through this without too much impact to our respective employees and families.
Barry, our customers are facing difficult times. Parodys and talking about futures does not help them today. Many bought a lot of capacity over the past few years that is currently under utilized for many reasons. Through storage virtualization we can provide the tools to help them recover stranded or underutilized storage and simplify management without replacing or buying additional storage capacity. Now is not the time to be pushing our customers to buy tons of new capacity even though it may help our revenues. Now is the time to help them be more efficient in the utilization of their current capacity, and provide them with the tools to scale rapidly when the economy returns.
I agree with you 100%.
But I don’t think customers want to suffer the tedious, time-consuming management forced on them by the USP-V, nor the performance impacts of virtualization, nor the wasted capacity of your Chubby Provisioning, nor the added expense of inserting another layer of hardware and network between their applications and their data. They can’t afford to invest in an aging half-solution in the back half of its lifecycle that locks them in with no future non-disruptive upgrade path, nor do they want to spend hours managing storage for their consolidated virtual server farms, or add yet more power-hungry heat-generating floor-tile-consuming equipment to their already bulging data centers.
No, the USP-V value proposition doesn’t even work on paper, much less in real life.
Customers want solutions that save them money TODAY – real money. For example, V-Max requires 20% less power/GB than a comparable DMX-4 with the same drives, and as much as 50%-70% less power/GB as older equipment with smaller disks. Couple the power savings with the unmatched performance, improved capacity density of Flash and SATA, simplified and automated management, extremely efficient thin provisioning, dynamic+fast+easy+nondistuptive reconfiguration, and customers are finding it CHEAPER to sweep out all the old gear and consolidate into DMX4 or V-Max than it would be to add your kit to what they have.
Symmetrix DMX4 and V-Max – BOTH are better, faster and easier than USP-V. BOTH use less power/GB and less power/IOPS than USP-V. And BOTH are tightly integrated with VMware, enabling even more efficiencies through server consolidation.
And BOTH are newer than your aged USP-V – few customers want to by yesterday’s technology when they can have tomorrow’s FOR LESS!!!
I see two different personalities in this thread.
Hu is a Priest and Barry (anarchist) is a Cow Boy.
Cow boys are tolerated during good times and priests are needed during difficult times.
I think we needs some good natured sermons from priest to help our customers rather than trying to bloat our profits in these diffcult times.
Barry, when are you becoming a priest?
To Counter Barry’s argument on low power consumption of V-MAX compared to DMX-4 here is my calculation which will prove it wrong.
V-MAX uses 24.2% more power per TB compared to DMX-4. EMC’s claims on using 20% less power per TB is false. These claims are absolutely outrageous about being green. May be DMX-4 is greener than V-MAX.
DMX-4 supports 1 System Bay and 8 Storage Bay. Maximum Scalability 2400 Drives(can be flash, FC or SATA)
As per my earlier calculation of power consumption
1 x System Bay * 6.4kVA = 6.4kVA
8 x Storage Bay * 6.1kVA = 48.8kVA
Total DMX-4 Power Consumption for 2400 drives = 55.2kVA
V-Max supports 1 System Bay and 11 Storage Bay. Maximum Scalability 2400 Drives(can be flash, FC or SATA)
1 x System Bay * 7.8kVA = 7.8kVA
10 x Storage Bay * 6.1kVA = 61.1kVA
Total V-MAX Power Consumption for 2400 drives = 68.9kVA
Barry, Any thoughts???
I think you need to read a little closer -
1. The DMX4 also supports 1 system bay and 10 drive bays, as does the V-Max
2. The V-MAx Drive Bay and the DMX4 drive bay use the same power, since they are in fact the same drive bays (and drives)
And of course, you have made a common mistake – you cannot compare power used by the Peaks Rated power ratings from the spec sheets. The actual power will be far lower – rated power is the maximum (surge) power required, rounded up to the next higher breaker size for the target geography. Further, each bay requires 2 power drops, and must be able to run off of only 1 in the event of a power loss to one source. So each feed must be rated at least 2x the nominal power draw.
I could stop there, demonstrating that you truly have not taken the time to understand power efficiency.
But I will also note there are additional factors in comparing power efficiency:
1) In the smaller configs, a V-Max SE can support more usable capacity than a DMX4-950 because of the V-Max new memory management architecture which delivers more usable capacity from the same amount of memory. More usable capacity @ equal or better performance + less power = greater efficiency
2) In its largest (11-bay) configuration, V-Max can support over 2PB of USABLE capacity. It will take multiple DMX4′s (or USP-V’s) to get to that same usable capacity – so even if the # of drives were the same, V-Max can support the capacity with only a single system bay, while DMX4/USP-V would require 2 or 3 (depending upon configuation)
Finally, all the Symmetrix power ratings include the power required to charge the integrated standby batteries – for an apples-to-apples competitive comparison, you have to include same for competitor products, whether the UPS are internal or external to the array.
Thanks, Ramash, for the opportunity to explain the superior power efficiency of both Symmetrix DMX and V-Max.
Wonderful explanation from what you like to say rather than what we/customers would like to hear.
First Common Sense
If I am customer and i am buying 2400 drive V-Max (oops!!!) first thing i will plan for is maximum power and weight in my data center. Your specs say that you use around 20% more power (curse..me!!! if you don’ like it) than a DMX-4 which you are accepting as a maximum that you will require. HDS USPV is more greener than your V-MAX in that sense. If you have an exact power consumption number you can share that information with your buddies here.
Point#1 DMx-4 1System bay and 8 Storage bay. Thats what your marketing says in the below link.
I also don’t agree to the fact that more usable capacity equals better performance. Being an engineer I know this is an ambiguous statement. If you want some help HDS engineers are willing to help you understand the same.
Last but not the least!!!
I know DMX-4 in full configuration (never configurable) weighs a staggering 13 tons concentrated over 10 to 15 Floor tiles. 95% of world’s data centers are not designed to handle that weight. Since V-Max also uses the same disk frames you cannot push the big V-Max whales into data centers eventually creating the same DMX-4 islands with a new name!! Old Wine in a new bottle!!! but not enjoyable!!
I appreciate you telling the real truth about V-Max and DMX-4.
You apparently have zero knowledge about power ratings vs. actual utilization. If you did, you would know that power ratings on spec sheets cannot be used to compare the power efficiency of two systems, since governing agencies require ratings to be the maximuum possible rather than “nominal.”
And though it may be a surprise to you, customers actually do buy storage based on usable capacity and performance, not drive counts.
The sad thing is you appear to not even know what it is that you don’t know. Or, perhaps you understand it quite well, yet choose to propose an abjectly illogical argument in hopes of confusing customers.
Either way, your responses are an insult to the intelligence of your audience. Fits in quite well with the other falsehoods and FUD found here on HDS.com, though.
I am sincerely surprised that you are allowing this sort of misinformation to take place on your blog, even if from a fellow HDS employee. After months of not engaging in the conversation yourself, this new approach to blogging engagement is just as meaningless as was your silence.
It’s your blog and you can do what you like, but approving this abjectly false information serves only to publicly tarnish your own reputation.
This has been an enjoyable toss about, thanks! I would like to add my (2) cents. I grew up on storage, my father was an STK engineer that got me into the business.I learned quite a bit from the STK folks, and IBM, they innovated! Fact is innovation in storage is at a stand still. Both of you high end vendors have created the current customer disposition with your glutinous overselling of product. When NetApp gets the Ontap convergance completed, the grid with global namespace will be king. I say that not as a NetApp fan, but out of first hand knowledge of their vision and direction, innovative. I am assuming HDS and EMC have similar visions, based on some of the aforementioned comments. But you are not there yet either, therefore you hype green data centers, which is laughable, and your software capabilities to manage the mess you have designed and sold, again laughable. Concerning software,HDS has made some recent announcements concerning new management tools, Aptare, that is confusing.Why would I buy it from HDS at an uplift and have my service passed through to Aptare? I know you will not be servicing it, you are in the process of outsourcing your backline support? Sorry for digressing, but if you folks could have an out of body eperience and take a clear view of yourselfs, there would be some head scratching. Some closing advice, build a product that is a grid,cloud is coming and you know it, align yourself in the stack with a server vendor, and HDS please get off of the virtualization pitch, VMWARE has and will continue to obsolete your big honkin bloated overpriced mega box capabilities.