Storagezilla

We don't do "free" Frank.

On the Data Domain earnings call CEO Frank Slootman bristled at suggestions that EMC had put the boot into Data Domain's margins.

Let's read the transcript.. 

Shebly Seyrafi

I’m wondering besides the macro situation what’s the say the second biggest negative swing factor that you saw recently? I wonder if it was EMC? For example, EMC talked about a $360 million annual run rate meaning $90 million per quarter, you were $85. I think they grew around 25% to 30% sequentially, you grew 14%. Maybe you can talk about how much of a factor this new competition with EMC was on your margins and your growth forecast.

Frank Slootman

First of all the number that EMC released I’m not too sure exactly what that number represents. It certainly does not represent disk library sent or shipped with deduplication abilities. EMC has for some period of time been doing a number in that order in their straight disk library EDL business with no deduplication whatsoever. Based on what we’ve seen during the quarter first of all we see that product being given away to customers. Certainly we’re in the [inaudible] it is free or near free so that doesn’t amount to a lot of revenue.

Then, our win rate against EMC has been very, very strong. The hard part about a company our size is we’re not in every deal and EMC [inaudible] in size [inaudible] by somebody like ourselves. When we had zero revenue EMC was already booking $60 million a quarter in disk library business so that puts things in to context for you.

Sorry Frank, to put things into context for you the $90M was just de-dup products. Avamar and the DL 1500/3000 series. The DL 4000 revenue is not included in that $90M. The 30% quarter over quarter growth was all de-dup.

As for the non-sensical point of giving away Disk Libraries for free since I sell them for a living I can assure you there isn't a CLARiiON array, DAEs, powerful Intel based server and contractual royalty payment to Quantum ending up in customers hands without EMC getting it's margin.

No way, no how.

As of writing Data Domain's stock has fallen by 32.41% while EMC partner Quantum has risen by 35.14%.

The EMC Effect is a rising tide for it's partners. Companies with useful technology should always keep that in mind.

The trouble with comments

Inspired by Marc's latest I want to talk a bit about comments. When TypePad finally added a proper anti-spam technology I ditched Captcha, it's a pain in the ass and half of the time when replying I couldn't make out a letter or two.

If the spammers come up with better Captcha breaking technology I'd appreciate a copy just so I can leave comments on TypePad blogs with Captcha enabled.

Eventually the flaphead contingent got used to someone working at EMC kicking around out on the Internet, the sock puppets working at various competitors (Well one in particular) got bored of me ridiculing them while the plain offensive weirdos just gave up and went home after turning my moderation queue into a cesspool.

So then I turned comment moderation off entirely.

Alas I erred when my fondness for new stuff over stuff which works got the better of me. Looking at TypePad Connect it had a few features I found interesting so I said I'd give it a go, moderation tools are far more limited in it but that was no longer a problem. I installed it, it ran fine apart some weird issues where you have to hit reload on the comment pages but otherwise worked well.

Comments were being automatically posted and I was happy. I went off on holiday, got some comments which I assumed were automatically posted and would reply to when I got back.

The problem was that while I was away something in Connect appeared to have changed. Even though comment moderation is disabled it still refuses to auto publish comments from users who have not signed in with a TypePad ID.

Which is super lame.

But on the upside it threads conversations and allows for email replies. Which is quite good.

Beta software is beta software so we'll see where they are when they deliver a onedotoh.

The NetWorker Blog

Independent consultant, author and NetWorker power user Preston de Guise has his own blog focused on EMC NetWorker.

We all think that's pretty cool and look forward to reading his insights.

And I like the URL in joke Preston. Clever. :-)

This Blogging thing doesn't matter *that* much.

Yes, yes this Internet thing is a mosh pit. Sometimes a mosh pit of competing ideas and that's when the roaring begins but for goodness sake never let it spill over into anything like this.

Mike Arrington is an arrogant so and so but what's even worse is that he's rude in person, that being said it doesn't justify someone spitting on him in the street. I'm sure there's a "He spits on enough people online and is a jerk in person" retort but following that line of reasoning to it's ultimate conclusion we're all still hanging out in trees and flinging dung at one another. 

There's a civility required when dealing with someone in person who you may not, and will never, see eye to eye with. I'm not saying the stor-o-spehere doesn't have it's share of contrarian buffoons, egomanics, fanatics and other toxic personalities who'll get you screaming, or in my case smiling as I know I'm going to needle them until one of us eventually dies, but be civil when you're out in the big blue room kids.

Today I was sure to shake the hand of competitors when given the opportunity as it's the civil thing to do. You'd be surprised at how many people in competitive situations don't. Though it is true that I did later put their personal details up on Craig's List in the "People seeking goats for fun/relationships" section for the world to see, but hey I want to win as badly as they do and I'm just willing to go further to make it happen...

Dude from Citrix trousers my clicker

How rude!

So there I was at a Server Virtualisation event earlier today where I go through how we're using VMware in EMC's IT organisation. We go through the adoption curve from POCs to your first Virtual Rack to your Virtual Data Center and the cost savings we've made with it.

Bottom line: This is the year EMC won't have a single x86 workload which isn't running in it's Virtual Infrastructure. 

From this year onward the only thing running on Intel based servers will be ESX.

I spoke about some of the different EMC technologies we're using, our building block approach, what's hot (Avamar had yet another barn busting quarter proving that Avamar & VMware go together like Strawberrys & Cream on a great Summer day), and what we think will be hot in the future.

I finish up and hand my clicker to the Citrix guy who earlier asked to borrow it. He does his thing (Very interesting to watch as it was really focused on application delivery more than Server Virtualisation/VM management with XEN) we do a quick Q&A.

And that was the last I saw of my bullet point shooting iron.

May all his PowerPoint presentations end in a BSOD. ;-p

EMC Q4 08: From the inside out.

Not a bad quarter, not great by any stretch since we missed our $15B revenue target by $100M but all in not rotten.

I was here in 2001 when we made a loss of over $100M and had to cut 29% of the workforce or find outselves out of business. We had an empty product pipeline, pricing which was 50 to 100% more than competitors and a single high revenue product which sales knew how to sell (Symm 5.5) built on a bus based architecture which had reached a development dead end.

We all knew it. Someone had difficulty facing up to the realities of the competative landscape and it resulted in his exile while the rest of us built a future without him.

This time around the new product pipeline is full. The Clariion line has recently been refreshed, EMC Unified Storage sales grew at more than 40% in 2008 and continue to deliver net new customers as well as competitor takeouts (The market share transfer continues), while other platforms will be refreshed this year.

And there's that mountain of money in the bank. 

At the last quarterly meeting having gone through where the market and EMC were headed Joe presented the company with seven directives. He listed them in minor detail on today's earnings call but I was glad to see that after a few quarters of inactivity M&A is back on the menu. 

Make room for PacMan..

RecoverPoint/Cluster Enabler (RecoverPoint/CE)

I was reading Brian's entry which bounced me over to John's blog where I remembered his entry on Cluster Enabler updates from November which caught my eye at the time.

I'm thinking when he said..

This past month, EMC has released a new version of SRDF/CE, MV/CE and internally announced a new addition to the Cluster Enabler family. I don’t think it’s been publicly announced, so I’ll have to refrain from posting any details on the third item at this time.

..he was talking about RecoverPoint/CE which was announced after his entry was written.

My corner of the universe has changed recently and RecoverPoint is back on my agenda. I think a corner stone of EMC's Information Infrastructure division has always been in replication technologies and as partisan as this most certainly will read (Apply filters accordingly) SRDF has to be recognized as King of the Mountain in the high end array based replication market while with more than 300K Clariions sold MirrorView is one of if not the most widely used array based block storage replication technology in the modular storage market.

Right in between them sits the network based RecoverPoint product line. Being heterogeneous means customers who may not have any EMC storage hardware can leverage the company's ongoing R&D expenditure in the area of replication technology without ditching your current storage provider(s).

For example, RecoverPoint/Cluster Enabler supports both heterogeneous and homogenous configurations. Which means you can mix and match third party arrays so long as they're on the RecoverPoint support matrix. This differs from SRDF/CE and MV/CE which are homogenous (Symm to Symm or CX to CX) only.  

One of the things I've always liked about RecoverPoint is that it's a treasure trove of interesting technologies which have applications in a lot of different areas. One of the coolest advancements to occur while I was busy with other things? Cascaded RecoverPoint. It provides three site disaster restart with consistent point in time images.

Groovy.

Your virtual money, gone in a FLASH.

EVE Online is one of the few single universe Massively Multiplayer Online games in the market place. It has ridiculously complex menus which only an engineer could love and incredibly dull combat even for an online game. But it features ships blowing up in space so I was sold on it within seconds.  

Being a single player universe and having so many concurrent connections to the same resources back in 2005 they chose to deploy a TMS RAMSAN (I think it was a 64GB unit at the time, I know it wasn't that much) in order to take parts of the DB off the performance pig of an IBM DS4300 and get things moving. Before that there was much cursing the crappy response times at peak hours of the evening. 

They later moved to a large server cluster which they've tuned to within an inch of it's life to run their game and replaced the internal HDs with SSDs. When FLASH as disk fails you've lost a disk. When FLASH on a card in a node or controller fails you've probably lost that node or controller and have to move the workload.

So by taking the logical step of using FLASH as fast disk, and not trying to beat it into being a cheap low performance cache RAM substitute, their database stopped being the bottleneck and life became better for all of us. Indeed it appears to have gotten a hell of a lot better for some since I read earlier that a player in a banking consortium made off with 80 Billion in EVE Currency and nearly took down a large player controlled banking establishment in the process.

The player's parting words? 

"So long and thanks for all the fish."

Xyratex is looking really ill

Xyratex (A supplier to Data Domain and others) is looking sickly right now.

So sickly that even I, who usually bounces over supplier news, have spotted the Angel of Death circling.

I'm currently looking at their most recent set of numbers and the prognosis isn't good.

Thank you SnapView

Having spent years as an Admin I still think like one and while that can be a hindrance when you're supposed to buy into some grand vision of the future my employer or a customer is pitching, some times it's a boon.

For example, this evening a frantic after hours phone call from a junior DBA I know at another company. He knows I'll still be at work and since he's in CYA mode he's not willing to start calling people at home so they can tell him he's an idiot.

"I've just made (expletive) of one of the databases we're putting into production and I'll be here all night if I have to set it up again."

"Okay. You expect me to do what?" says I. A typical admin answer derived from a long day and one eye on the clock even when you're not going anywhere.

"Our Storage Admin said he was going to take snapshots could we restore from those? I have the login password."

And then I relax.

He logs into Navi Manager, we find SnapView sessions for the database volumes and I tell him to shutdown the running database. We flush the buffers and I let him know we'll be keeping a Recovery Session of the current state.

"But it's corrupted, why am I doing that?"

"You're doing that so when your boss comes in in the morning and asks you what happened not only can you tell her but you can mount it on a different host and show her if she asks to see it. It's also a sanity check in case a couple of hours from now you're sitting at home and realize that in a fit of panic you made the wrong call."

We click our way through rolling back the consistency group, and bring everything up. The DB comes online but is missing some of his later changes so he'd like to roll forward to one of the later snapshots.

And that's where with some other snap technologies on the market he'd have been screwed. To some vendors snap rollback is a dead end.

Go back too far and you end up trapped in a cul-de-sac on the road to greatness. Everything which came before it consigned to oblivion.

So we roll forward and he's happy with the result. I warn him to write up an accurate post mortem before he goes home and mail it to his colleges before the storage admin spots activity in the log files and starts asking questions.

But he can omit the fact that I was on the other end of a phone line making sure he was doing the right thing.

Screwing up and fixing it is nothing to be ashamed of. I've sat in front of Sun boxes having been up 36 hours straight, surrounded by the detritus of empty Pepsi tins and sweet wrappers, only to run a set of commands which my tired brain, operating a split second behind my fingers, realized I had just typed into the wrong SSH window.

Knowing how to walk back out of it meant the only person cursing me was myself.

So what have we learned from tonight?

1. The Storage Admin did their job. SnapView was configured and running on those LUNs well before tonight's adventure began.
2. Before you perform any destructive changes be sure to take a snapshot first even when you think you're dealing with nothing but garbage. Screw ups tend to get orders of magnitude worse when you're operating under blind panic.
3. Not all snapshot technologies are the same. Be sure you don't lose all of today if you decide to take a quick trip back to yesterday. Today matters. The previous hour matters. 

And finally, even when someone is hyperventilating down the phone..

      4. Zilla's still got it.