The Storage Architect

Enterprise Computing: Run My Storage At 60%? No Way!

Chris Evans — Wed, 17 Mar 2010 19:52:53 +0000

Hu Yoshida has an interesting view on his recent post discussing storage utilisation rates. His concluding remark suggests running at a maximum of 60% utilisation – even with Dynamic Provisioning. Hu, you must be joking, right?

Point 1: I’ve paid for my 100% of storage and I’m going to use it. I don’t remember any vendor suggesting only paying 60% of their invoice and calling it quits. Granted, spending an inordinate amount of time to reach the 80%+ goal isn’t necessarily cost effective, however it can be achieved.

Point 2: It’s perfectly possible to run at 80% utilisation. It just takes some thought and planning. As Hu rightly points out, Dynamic Provisioning helps significantly towards this. The features of wide striping and thin provisioning mean storage can be more easily provisioned and requires less manual balancing.

Point 3: Achieving high utilisation isn’t all about technology. It’s also about process. That means Demand Planning, Capacity Planning, efficient processes for deployment of new hardware and for provisioning of customer requests. All this is possible without incurring additional expense.

So, don’t think 80%+ isn’t an achievable target. After all, it’s your money you’re wasting if you don’t try!

Cloud Computing: Cloud /= Virtualisation

Chris Evans — Mon, 15 Mar 2010 13:04:13 +0000

I finally managed to attend a London CloudCamp last Thursday, which conveniently co-incided with a #storagebeers evening. For two hours of listening to the collective wisdom of the presenters and the “unpanel” we were offered free beer and food. Now free beer is good, however I’m not sure it’s good enough to make me want to listen to yet another presenter confuse Cloud with Virtualisation. This may have been a genuine misunderstanding, however it may also have been quite deliberate, as the speaker in question works for a VMware partner. However moving on, here’s what I think Cloud is, should be and will be.

My Cloud View

Cloud is not virtualisation. Virtualisation is virtualisation. It’s a way of consolidating the physical entities we know today and making them available in the same manner in a virtualised form. Moving 1000 Windows servers to 1000 virtual machines changes nothing. You’re still offering IT to customers in the same way – using the concept of that physical entity called a server. This is a step forward in reducing cost, but not in delivering cloud.

Cloud today is a hybrid of two things; virtualisation & services and service abstraction. In the first form, Amazon, Rackspace and the others are offering the same physical objects (i.e. servers) but packaged nicely so they can charge you efficiently for them. What makes this kind of environment a step above virtualisation is that you have no view, concern or interest in how, where or by whom those servers are being provided as long as they meet your requirements for security and availability. The hosting company is effectively offering you elasticity over and above the basic virtualisation offering you could achieve yourself. If you need more compute, then they guarantee to provide it and they manage the pain of ensuring resources are available. Second, is service abstraction; some things are non-server dependent. For instance; generic databases or email outsourcing. The hosting company offers you a service, rather than a virtual device. This leads on to the next step; what Cloud will be in the future.

To see what Cloud Computing can offer, we need to move forward away from the fixed view of infrastructure and start thinking about data and applications. In the future I want to move away from thinking about fixed data types like files and block devices and think about objects and metadata. These objects need to be modifiable in a way that gives me consistency, so if I lose a node, or device that’s processing the data, I can recover it from elsewhere and continue working. I want to process my data in parallel and without geographic constraints, so I can run or move my workload anywhere; presumably to the provider who is offering me best price on that day.

The treatment of data in this form means I need to rethink how applications are written. I can’t rely on a simple client-server hierarchical architecture but need to move to one where my processing and data is distributed.

Re-writing distributed applications will be the hardest part of delivering Cloud Computing.

Application development in a completely distributed environment creates significant challenges. Here are a few of the most obvious issues;

How do I ensure consistency (multiple updates in multiple locations)?
How can I manage performance?
How can I move my data around easily and quickly?
How can I back my data up?
How can I move my application and/or data to make best use of cost savings?
How can I ensure my data remains secure?

The ultimate Cloud Computing model needs to remove almost all the aspects of infrastructure structure we know today (server types, storage arrays, block devices, NAS) from the customer. It will offer abstracted programming concepts – store, retrieve, search, update – to process data. Customers will be charged for their consumption of these resources. Abstraction will allow applications to be run on (almost) any environment. These services will still be delivered on infrastructure, but we won’t care or need to care what it looks like. Now there’s a challenge…

Enterprise Computing: COPAN, EMC/VMware & STEC

Chris Evans — Thu, 11 Mar 2010 10:00:39 +0000

Over the last week there have been a few stories catching my eye. Here’s a brief paragraph on them.

SGI Acquires COPAN Systems

In fact to be more precise, SGI have acquired some of the assets of COPAN and left the liabilities behind for a mere $2 million in cash (press release). The demise of COPAN raises two potential questions; is spin-down a dead technology or were COPAN in a market that wasn’t able to understand their technology? I have to admit it took me a while to get what COPAN were offering. In fact I had a presentation of their technology by the UK team a couple of years ago in which I experienced that “light bulb” moment. However times change. Ask yourself the following questions:

How expensive was COPAN compared to a standard modular solution?
What savings could I actually achieve in a COPAN solution?
Which other vendors now offer spin-down in their products?

At the end of the day, the majority of an array is disk drives, which cost the same, whoever you purchase them from. Savings from suspending power usage are not that great; powering down a single drive may only save $10-15/year, hardly a massive saving. With other vendors also offering spin down in their products, it was only a matter of time before COPAN hit a wall. I evaluated the product for deployment at a client. My major issue with the offering was the sheer weight of it. This factor alone would have made impractical for many data centres.

EMC Sells IP to VMware

It’s a bit like borrowing something from your parents; EMC are selling some of their software IP to VMware for the princely sum of $200 million (press release). The products being transferred relate to server and application management. Does this mean that owning the Hypervisor isn’t enough to beat the competition? I expect it does. Where’s the money to be made in selling a hypvervisor that in a few years will have to be given away for free? Hyper-V may not be competitive today, but a free product definitely beats a one that has to be paid for, even if the features aren’t quite as good. So the future is selling add-on value, simplification or tie-in. Maybe that’s the reason behind the parental selloff.

STEC Shares Get Punished

STEC shares have yet again taken a battering after they missed earnings target by a mile (Barrons Report). The press report on the link makes an interesting comment claiming EMC may have effectively stopped ordering SSDs.

I’ve personally always thought that SSDs were overhyped. I never believed they would be adopted as quickly as EMC predicted (or more accurately would have wished). Have a look at Beth Pariseau’s article from this time last year (TechTarget article). At the time Tucci claimed EMC were selling “every SSD they could get their hands on”. We can only assume that has changed. What a difference a year makes.

Enterprise Computing: HP Blades Tech Day – Roundup

Chris Evans — Wed, 10 Mar 2010 06:49:31 +0000

Here’s my roundup of all the posts, pictures, video and comments from the HP Blades Tech Day

Tweets

The official hashtag for the event was #hpbladesday with hundreds of tweets being generated from the start of the Tech Day until now.

Bloggers’ Posts

By name order, here are the relevant posts from each blogger.

Kevin Houston (http://www.bladesmadesimple.com) - @kevin_houston
Greg Knieriemen (http://www.storagemonkeys.com) - @knieriemen
- Interview with Gary Thome about HP Blade Servers

Frank Owen (http://www.techvirtuoso.com) - @fowen

John Obeto (http://www.absolutelywindows.com) - @johnobeto

Rich Brambley (http://www.vmetc.com) - @rbrambley

Simon Seagrave (http://www.techhead.co.uk) - @Kiwi_Si

Martin MacLeod (http://www.bladewatch.com) - @martinmacleod

Devang Panchigar (http://www.storagenerve.com) - @storagenerve

Stephen Foskett (http://blog.fosketts.net) - @sfoskett
- Innocence, Fairness, and Technology Benchmarks

Chris Evans (http://www.thestoragearchitect.com) – @chrismevans

Presenters, Presentation Material, PR

Present from HP and blogging during the event:

HP presenters were:

Gary Thome – Chief Architect, Infrastructure Software and Blades (ISB), HP
Mike Kendall, Manager, HP Virtual Connect Marketing, ISB, HP
Mike Witkowski, Distinguished Strategist, ISB, HP
Mark Gonzalez, Nth Generation Computing
John Schmitz, Product Marketing Team Leader, ISB, HP
Bryan Jacquot, Software Architect, ISB, HP
Tom Turicchi, Engineering, ISB, HP
Joseph George, Client Virtualisation Business Lead, ISB, HP

Managing the Factory Express Tour were:

Denise Herdman, Supply Chain Operations Business Development Manager, Factory Express
Alex Irvine, Director of Supply Chain Operations, Factory Express

PR and herding the blogger cats was managed by Ivy WorldWide: Halley Bass and Tom Augenthaler did a fantastic job, as did the PR team from HP; Becca Taylor and Jason Treu.

Sincere apologies if I’ve missed anyone off; let me know and I’ll add them. It was a great event, there is still more content from me to come.

Virtualisation: Home Lab Update

Chris Evans — Tue, 09 Mar 2010 07:28:59 +0000

Home lab storage

I’ve just about got my home lab fully organised; here’s a quick summary of what I have in place.

Storage

I’ve four main storage devices, DroboPro, Iomega ix4-200D (two) and an EMC CLARiiON CX-700. This gives me NAS, iSCSI and Fibre Channel connectivity options on any testing I’m performing. The storage is broken down as follows:

DroboPro – 8.4TB of raw storage across 7 disks (nearly fully populated) – iSCSI
Iomega ix4 – 2x 4TB of raw storage across 4 disks – iSCSI & NAS
CLARiiON – 2.2TB of raw storage across 15 disks – Fibre Channel

This gives me 18.6TB of raw capacity. One of the ix4 devices is production, running my VMs on ESX, the other is for testing performance and configurations but will ultimately form my remote backup solution.

Servers

Server Rack

My main server is an Intel Based, 2 CPU, 8 core custom-build machine with 16GB of RAM. It runs ESXi 4.0 and has access to all storage resources as well as another 1.6TB of internal storage. This machine runs all my production and test virtual machines. A second server runs Windows Server 2008 R2 with Hyper-V (upgraded this week). This is my Hyper-V test server. It has a single 500GB drive and access to all storage resources, including fibre channel via an Emulex LP12000 HBA card.

I also have two other smaller servers that act as test machines in various scenarios, one of which is currently acting as an OPEN-E server (the subject of a future post).

Fabric

The fabric in my test lab is provided by a number of McDATA 3232 switches. Although they are servicable, they are looking old now, so I’m looking to acquire a Cisco MDS switch to add to these for some time in the future. Having said that, the McDATA switches do just work and are a perfect example of technology that doesn’t need to be replaced.

My setup allows me to test and experiment with a wide range of virtualisation scenarios. Obviously I’m reliant on free ESXi software so I can’t always test the features I’d like such as vMotion unless I continually re-install ESXi. However it does give me an opportunity to understand the operational issues of vSphere.

Next on the testing agenda is looking at Hyper-V. The Windows 2008 Hyper-V server I have in place has been upgraded to R2 and I’ve started to create a number of test environments, including testing with DroboPro.

Stay tuned for more updates!

Enterprise Computing: HP Blades Tech Day 2

Chris Evans — Fri, 05 Mar 2010 06:48:39 +0000

Day 2 of HP Blades Day took place in a new location – the Customer Experience Centre. This was due to issues with Wifi on day 1, now resolved in our new location. Order of the day was:

Review of the previous day’s presentations with a quick question & answer session
Presentation on Client Virtualisation
Factory tour

Client Virtualisation

Joseph George presented to the bloggers on client (aka desktop) virtualisation. Whilst HP seem to have a story in this area, I’m skeptical about the whole process of virtualising desktops at this stage, other than in certain use cases. I can see benefits for the following:

High availability environments such as financial traders, where loss of a desktop translates to financial loss.
Large scale desktop deployments where functionality is generic. A good example of this is call centres; desks only require access to limited features (so don’t need high powered devices) and physical desktops may be used by multiple users.

Environments not suitable for desktop virtualisation will include mobile users and anyone running bespoke hardware or software with hardware dependencies.

HP pretty much accepted that hardware savings are not the main reason for virtualising. This means a more intelligent model on TCO for desktops needs to be established, encompassing consequential loss through hardware downtime, processes cost for technology deployment and replacement and so on. I think some of the thinking around security is a red herring as laptops are unlikely to be candidates for virtualisation and desktop data can be secured safely today. I’ll be writing more on this topic over the coming weeks.

Factory Tour

We next moved on to a tour of the HP factory; this isn’t a manufacturing plant but rather purely an assembly line for server and storage components. I’ve been on factory tours before and these days they pretty much leave me cold. However HP had some interesting innovations to discuss.

Customisation. All HP server and storage builds are based on customer requests and so fully customised, including placing of kit into racks for shipping. This means every configuration shipped is exactly as the customer ordered – and more; HP also pre-cable the racks as the equipment is being installed. Customers ordering multiple configurations can preview one rack of their order before the remainder are built.
Testing. Once servers are built, customers are provided the ability to VPN into their equipment while it is sitting on the factory floor. They can perform load and performance testing before the equipment ships, saving valuable deployment time once the equipment arrives.
POD (performance optimised datacentre). This is HPs term for a datacentre in a container – which is exactly what they are. Plug in power and cooling and you’ve a data centre on the move. Except these devices aren’t really built to be picked up and moved around. Rather they are aimed at providing temporary data centre facilities in scenarios where traditional data centres wouldn’t be possible – extension of existing facilities or perhaps creation of a data centre without planning permission were two good examples.

After the tour, we wrapped up at a local restaurant.

Overall the two days have been extremely informative and given me many subjects to blog about. Look forward to more over the coming weeks!

Enterprise Computing: Sun/Oracle Kicks Hitachi To The Kerb

Chris Evans — Tue, 02 Mar 2010 22:16:18 +0000

I’ve just been reading Greg Knieriemen’s post over at iknerd.com on the ending of Sun/Oracle’s OEM agreement with Hitachi to resell their high end storage arrays. On the one hand I’m surprised by the announcement; on the other I’m not.

Sun have resold Hitachi for some time under the 9990V and 9985V brand names. These are no more than rebadged devices with Hitachi code and software under the covers. From memory, I believe the only technical change is the cabinet door. I have installed Sun supplied Hitachi technology in the past. Sun provided no added value to the process – in fact when we encountered a microcode bug, Sun’s lack of knowledge hindered our problem resolution process.

I’m struggling to see what Sun/Oracle will replace the Hitachi brands with. All of the original StorageTek disk array products are effectively dead. Sun’s Unified Storage Server (7000 series) is a great box but not enterprise class, neither are Sun/Oracle’s other offerings, all based around modular storage. I can see three options:

Larry doesn’t want to replace Hitachi’s products. They weren’t owned by Oracle so didn’t suit his plans, so they’re gone.
Use Exadata. OK, that doesn’t give the customer much flexibility in their datacentre.
Find a new partner. Perhaps after all the hype we hear about Netapp in Oracle’s datacentres this is the time for the switch. Maybe we are going to see Netapp as the preferred enterprise storage partner.

To be honest, no option seems likely. I guess time will tell.

Oh, nearly forgot, there is of course the Pillar option, but I don’t see that one happening either.

Enterprise Computing: HP Blades Tech Day 1

Chris Evans — Tue, 02 Mar 2010 11:51:40 +0000

HP Blades Tech Day 1 is now over and what a whirlwind of a day it was. Here’s a summary of the day’s activities:

Overview of ISB (Infrastructure Software and Blades) from Gary Thome
Partner Presentation from Mark Gonzalez
Overview of Virtual Connect from Mike Kendall
BladeSystem lab (hands on overview of the hardware) from Gary Thome
BladeSystem Matrix & Insight Software demo from John Schmitz & Bryan Jacquot
Competitive discussion (hardware) with Gary Thome
Face-to-face discussion on strategy with Paul Perez (CTO for StorageWorks)

I can honestly say that there wasn’t a bad presentation all day. We were given access to some senior technical people (for instance Gary is Chief Architect for ISB) and that showed – as we were all able to ask probing questions and get immediate answers. For me, a number of things stood out for further investigation.

The blade chassis themselves have a huge amount of engineering in them. I’ve taken a lot of video of the lab presentation which talks about the power supplies, fans, physical layout and so on. It will take time for me to review and digest but expect some posts on the engineering that has gone into the blade products.
Power Management – this subject deserves a post in its own right as there’s some clever thinking here.
Virtual Connect – virtualisation of the I/O infrastructure in any blade system is a key requirement and VC had some interesting features. I’ll be looking to compare VC to other solutions in the marketplace.
Software and Storage – I don’t think storage has been fully exploited in the current solutions. Rather it is seen as a static configuration achieved separately and doesn’t form part of the workflow in the dynamic deployment of infrastructure. Look out for a post discussing this whole issue.
Futures. Although Paul Perez didn’t explicitly provide futures information, he hinted at the direction HP may take for the storage products. Expect to see a post on memristors and the future direction of hard drives.

In summary, day 1 was a great introduction (for me at least) to HP blade technology. HP were open and willing to discuss their technology and I can say I learned a lot, with plenty of research required. Now onto day 2….

Disclosure: HP have funded my flight and accommodation for this trip. Everything else is self funded. I am under no obligation to write about what I see during the Tech Day and HP have no editorial rights over the content I produce.

Enterprise Computing: HP Blades Tech Day 0

Chris Evans — Wed, 24 Feb 2010 22:55:51 +0000

Some of you may be aware that I’m heading to Houston today for an HP sponsored field day looking at blade technology. I have to admit that blades are not my area of speciality, in fact I know precious little about them other than the obvious. Having said that, this is a time to start being more aware of the technology “stack” as we move towards converged architectures in the data centre.

By convergence, I’m referring to what we already see with VMware, Cisco and EMC (VCE); the Acadia triumvirate. HP have already started to talk about their convergence strategy and I’m sure other vendors have ideas in plan. I’m looking forward to understanding where HP’s strategy is headed, how their relationships with VMware and Microsoft will play out at the virtualisation level and after this week, how they intend to manage the (storage) networking component.

Fortunately I’ll be in the company of some experts to keep me honest. Travelling with me from the UK is Simon Seagrave, plus the other following attendees:

Kevin Houston (http://www.bladesmadesimple.com) – @kevin_houston
Greg Knieriemen (http://www.storagemonkeys.com) – @knieriemen
Frank Owen (http://www.techvirtuoso.com) – @fowen
John Obeto (http://www.absolutelywindows.com) – @johnobeto
Rich Brambley (http://www.vmetc.com) – @rbrambley
Simon Seagrave (http://www.techhead.co.uk) – @Kiwi_Si
Martin MacLeod (http://www.bladewatch.com) – @martinmacleod
Devang Panchigar (http://www.storagenerve.com) – @storagenerve
Stephen Foskett (http://blog.fosketts.net) – @sfoskett

If you’re following on Twitter, the official hashtag is #hpbladesday and you can also follow @bladenews

Athough it’s only a three day visit, I’m hoping to see something of Houston; ten-gallon hats, steaks and firearms come to mind. More updates to come!

Enterprise Computing – Death of Tiering?

Chris Evans — Tue, 23 Feb 2010 15:44:43 +0000

I’m not the first to post on the subject of Netapp’s President and CEO Tom Georgens commenting during their latest earnings call on the apparent death of tiering as we know it today.

In Netapp’s view, there will be no tiering of storage in the future. Instead we will be using SATA drives for our data and cache cards. Here’s Tom’s words, taken from the call transcript:

Second of all, frankly I think the concept of tiering is dying. And I probably don’t want to go into a long speech on that, but at the end of the day, the simple fact of the matter is tiering is a way to manage migration of data between fiber-channel based systems and serial ATA-based systems. With the advent of Flash, and we talked about our performance acceleration module, basically these systems are going to large amounts of Flash, which are going to be dynamic with serial ATA behind them, and the whole concept of HSM and tiered storage is going to go away.

So what are Netapp thinking? I thought the PAM module was a read cache for accelerating random I/O read-intensive environments? How would that help scenarios where there’s heavy write activity? This is where FC and SSDs are most suited and Netapp are saying they won’t be needed?

Perhaps Tom’s comments preview a change to the Netapp architecture in which the PAM cards are used to improve write caching. That’s how Sun’s 7000 Unified Storage Systems work; use SSD for caching writes and serve data from SATA. Maybe SSDs are too fast for Netapp’s architecture and this is why they need to implemented via the back door.

Tiering today enables customers to match workload to the price of the underlying storage hardware. EMC and Compellent are most well known for introducing technologies that enable data to be moved at a granular level, making best use of SSD. No doubt the other major vendors will follow suit as they release the next generation of their products. Storage arrays that can move small blocks of data onto the most appropriate tier of storage will deliver the next wave of efficiency. After that we should expect the array to manage the underlying hardware automatically, while we specify policies that dictate the service we expect. Tiering isn’t going away any time soon.