|RSS | MY PROFILE | PRIVACY|
Vol. 17, #44 - October 29, 2012 - Issue #903
This week's issue of WServerNews is all about storage tiering and includes some upcoming plans for my own business in this area. But managing data storage isn't just about technology, it's also about implementing good business processes and practices. But even with the very best intentions, such processes and practices can often deteriorate over time until they START TO STINK:
The Tip of the Week in the previous issue Top Server Management Resources (Issue 902) generated a lot of reader feedback. For those that didn't read it, here's the tip again:
If you plan on deploying a bunch of servers with internal hardware RAID storage, it's better to buy all of the hard drives at the same time from a single vendor. That way all the drives will have the same firmware level which means they will less likely to have issues when used in RAID configurations. Plus your service and warranty management will be a lot easier.Of course, before you do this make sure you research the literature for reviews of the make/model of drive you plan on buying to ensure you're getting drives that have a track record of reliability and not a bunch of lemons.
A reader named Bruce responded with the following suggestion:
It may be obvious to some but when buying all of your raid drives at the same time, remember to enough replacement drives to replace those drives that may fail over the expected life of the servers.
Brett from Australia made another suggestion and also asked for some feecback as follows:
In reference to your section on Raid and buying the same drives for firmware conformity - you may be aware (and I was advised - now 3 years ago) that that many RAID set ups ALSO depend on the Mainboard (motherboard, m/b) set up as well. So it may also be useful to also adopt the same motherboards. That would also help with driver and other issues.
I run only 1 server with RAID 0, so I am no big shop as such, but the server and at least 3 other PCs use the same motherboard so if the RAID Server m/b fails I can transfer a m/b to the server and keep going. But I am no expert and would appreciate feedback on this!
Great Newsletter and I do keep it for reference!
What do readers think about Brett's suggestion concerning motherboards, and in general about bulk-buying and over-provisioning of computer hardware components? Send me your feedback at firstname.lastname@example.org
Calvin from Ontario, Canada, voiced a different perspective on bulk-buying HDDs like this from a single vendor:
If buying from the same lot you are now on the same MTBF (Mean Time Before Failure) for all drives. Despite looking for the best possible drives - once one fails in an array the possibility is very high for another to fail in that same array. I would think that even a hot spare would be at similar risk if it is a rotational issue (bum bearings). You could buy another spare drive and leave it on the shelf (what I do for my clients) or drive yourself crazy verifying lot numbers of the drives. FWIW - the last time I had a drive replaced by an OEM (that supplied the RAID setup) the replacement drive actually had an older firmware on it - and they would not do anything to change that (:<
I've heard that argument as well, but on the whole I tend to think the benefits outweigh the risks on this issue, at least when it comes to datacenter environments where you often do bulk purchasing.
Finally, Tony from the UK shared at length his approach to minimizing the impact of possible HDD failure as follows:
I am not so sure that the perceived conventional wisdom of getting all your RAID disks from the same batch is such a great idea. Many years ago, a company I dealt with had a server with RAID1 - a pair of SCSI disks. They both failed over the same weekend.
Back in about 2005, I built myself a new server with RAID1 and a hot spare - three Seagate SATA disks from the same batch. I could not understand all the problems I was having, but after a month, I deduced that all three were faulty.
A couple of years ago, I had an HP server for a customer supplied supposedly soak tested by a UK supplier who shall be nameless. When I tried to install Windows SBS2008, it kept crashing. Turned out the hardware was bad - the supposed soak test was limited to 30 mins.
I actually take a different approach. OK, I may end up starting from a pair of disks (RAID1) from the same batch (and I run them for a few weeks before taking them live to get through the infant mortality phase). Then after something like 18 months, I replace one of them. Usually by now it is a faster and bigger disk, so although different, it does not degrade the hardware RAID. 18 months later, I will replace the other one, and as this is faster again, I usually get a performance boost, and can extend the capacity to that of the smaller of the pair.
The reasoning behind this is that I now protect myself from infant mortality, and by replacing disks before they are approaching end of life (and I have seen less than 3 years for 24/7 disks in a NAS recently, and less than 1 year for their replacements), I reduce the risk of failure in close succession e.g. over the same weekend. After all, if you think about it, with modern manufacturing controlled so tightly, and disks being operated in the same environment (i.e. in RAID1 in the same chassis), the probability that they will fail closer together is actually increasing.
A few years ago there was a big study on disks that concluded that independent of disk type - SCSI, IDE, SATA, the real disk MTBF (Mean Time Before Failure) was 10x less than quoted by the manufacturers. This was in line with my observations over decades that disks were 10x less reliable than the manufacturers figures. Which is why I adopted my step by step replacement plan outlined above.
Just to round this out - the reliability of non-mechanical devices is related to the number of "pins" (external connections now we are in surface mount) and the thermal regime and mechanical stress. This is why ever greater integration is leading to higher reliability. It is obviously a lot more complicated than that, but as a first order approximation it is a reasonable guide. Mechanical devices are similar, but obviously there is wear and vibration. Again, two devices running for the same amount of time in the same environment should wear at the same rate and suffer the same mechanical conditions.
Thanks for all the terrific feedback. If any other readers would like weigh in on this topic, please feel free to email me at email@example.com
It's funny how businesses evolve over time, even small businesses. Apart from being Senior Editor of WServerNews, the world's largest Windows Server-focused newsletter, I also run an IT content development business based in Canada that produces books, whitepapers, articles, courseware, and other content that targets business and technical decision makers, system administrators, students/trainers, and helpdesk. And even though our business is small, it's been running almost 15 years now and our storage infrastructure has evolved like most businesses do, that is, organically into silos. What I mean is, whenever we've needed more storage we've usually just gone out and bought whatever seemed like the best deal at the time. The result of such lack of planning is that our server storage technologies are currently a combination of internal HDDs, external eSATA enclosures, network attached storage (NAS), external USB drives, memory sticks, Windows SkyDrive folders, and so on.
Pretty messy, eh?
We plan to change our approach by trying to think more like a large enterprise than a small business. As a key part of this rethinking process, we're creating a plan for storage consolidation and growth that will cover our needs for the next several years. Our goals are to make storage management easier while ensuring we can quickly find whatever data we need. We also want to improve the performance of some business-critical applications that rely heavily on access to both local and network storage. And we want to make sure everything is reliably backed up and can be restored when needed in a reasonable amount of time. Finally, we have to make sure our plan can be implemented at a reasonable cost to our business.
Defining data tiers
The key to accomplishing all this starts with data tiering. Planning storage management for large enterprises usually starts with defining what company data should be assigned to each of the following three tiers:
There's a close relationship between the type of data and the technology you should use to store it:
Considerations involved in planning for data tiering include:
What we're considering
We're currently taking a hard look at Storage Spaces, a new feature of Windows Server 2012 that lets you virtualize storage by creating pools from which you can then create virtual disks that can have mirror or parity resiliency. You can read an overview of this new technology here:
We're thinking we could use Storage Spaces to create two pools, one for provisioning storage for hot data and the other for provisioning storage for cool data. Our hot data pool could use storage devices like internal 15K HDDs and/or SSDs, while our cool data pool could use an external JBOD enclosure with high-capacity HDDs. We're currently waiting for the Windows Server Catalog to list more JOBDs certified for use with Storage Spaces:
So far, there's only one certified offering listed, which is this one from DataON Storage:
This 24-bay JBOD might be a bit more than we need (we've been thinking that a 4-bay JBOD might be sufficient for our needs) but we really won't know until we've actually quantified all our existing data and estimated for future growth.
As far as archival storage is concerned, we're currently looking at different cloud-based backup solutions but have not made any firm plans whether to go this way or not. Basically, we're taking our time as we want to "get it right" this time, so if it takes us until next summer to design and test the best solution, that's just fine with us.
Send us feedback
Have you implemented any form of storage tiering in your own business? Got any suggestions or recommendations to share? Email us at firstname.lastname@example.org
You have a server that supports Intelligent Platform Management Interface (IPMI) and you need to flash the BIOS on the server to update it. The vendor provides you with an .exe file for doing this and tells you to copy it to a floppy drive and boot from the floppy to flash the BIOS. Unfortunately the server doesn't have an internal floppy drive, and you don't have a USB floppy drive laying around. What can you do?
If the server is a Hyper-V host with a virtual machine running some version of Windows, you can use Hyper-V Manager to create a virtual floppy disk (.vfd file). You can then mount the virtual floppy disk by selecting it on the Diskette Drive page of Hyper-V Settings, boot the virtual machine, format the virtual floppy disk and copy the .exe file from the host to the virtual floppy disk in the virtual machine. Then shut down the virtual machine and use IPMI to redirect the floppy disk image to flash the BIOS.
And for information on why Hyper-V even bothers to include support for virtual floppy disks, see John Howard's blog post at
Got tips of your own that you'd like to share with our readers? Email me at email@example.com
Information Security Governance Simplified from CRC Press helps you implement a cost-effective security program for your organization that complies with government regulations. The book begins with an explanation of what information security governance is and then describes a process for defining the security management of your organization. Strategies for interaction between CEO, CIO and CISO are described next, followed by risk management. I found the section on risk management especially helpful as it dealt with underlying assumptions concerning risk and also risk avoidance and transference, which are not often distinguished properly. The rest of the book explains how to create effective infosec policies, how to comply with various frameworks like HIPPA and ISO 27001:2005, various controls (managerial, technical and operational), and how to deal with auditors and the law. Excellent book, well-written, and definitely recommended reading for infosec professionals of mid- to large-sized organizations.
"Use only that which works, and take it from any place you can find it" --Bruce Lee
Until next week,
SolarWinds trio of free tools helps you manage and remove inactive computers and users from Active Directory, and allows you to add users in bulk. Download your three free tools today!
For expert Hyper-V advice, guides, ebooks and more sign up to receive great content by email
Break through the barriers of Microsoft MMC tools and the complexity of PowerShell with the most powerful version yet of SystemTools Hyena - Download a free, fully functioning 30-day trial.
Test Drive the StealthAUDIT Management Platform for Exchange. SMP provides superior Exchange Auditing, Exchange Reporting, Analysis, and Remediation capabilities all wrapped into a single solution. Request A Free Trial.
EasyBCDis a very simple UI for bcdedit that makes dual-booting really easy:
Contact Michael Vella at firstname.lastname@example.org to get your conference or other event listed in our Events Calendar.
Learn what archiving functionality to expect in Exchange 2013. Should you use or update your third-party archiving solution? Join Osterman Research and GFI Software for a new, educational webinar. Drawing for new iPad for one registrant.
Contact Michael Vella at email@example.com to get your webcast listed in our Webcasts Calendar.
Cisco Web Security Appliances was selected the winner in the Antivirus Hardware category of the WindowsNetworking.com Readers' Choice Awards:
Two new online courses have just been released at Microsoft Virtual Academy:
Eric Geier discusses the new data backup and recovery features introduced in Windows 8 in this article on WindowsNetworking.com:
Windows Server 2012 Essentials (formerly Windows Small Business Server Essentials) is a server solution designed and priced for small businesses with up to 25 users and 50 devices:
And for some product highlights, see this post from the Windows Server Blog:
This public forum is a place for engineers testing and certifying Windows hardware, drivers, and systems to share knowledge, get questions answered, and learn from shared experiences:
Developing applications in the cloud can offer many advantages, but the process is not as straightforward as you might think. Luckily, there are a number of development tools you can leverage to streamline the process. Find out more in this essential tip.
Hyper-V live migration offers many advantages, but it is often a challenging task for IT pros as there are many factors that can go wrong. Inside this exclusive tip, review the top reasons why live migrations fail so you can ensure that you don't make the same mistakes.
While virtual desktops offer many benefits, the significant increase in VDI network traffic can introduce challenges if you don't plan ahead for it. Learn key tips that can help you effectively prepare your network infrastructure so you can take full advantage of your new virtual desktop environment.
When it comes to Windows 7 migration, IT professionals often face a number of challenges. Fortunately, there are a variety of advanced, built-in tools available that can simplify troubleshooting tasks. Access this exclusive resource to explore your options.
Ever wonder what Google's datacenter looks like from inside? Check out the photos in this article from Business Insider:
This up-tempo piece from the second Animusic DVD features a band of five robots jamming on their futuristic instruments as their musical starship cruises through outer space:
The best skiers, surfers, divers, bikers, kayakers and pilots filmed with the GoPro Hero3 camera.
Fantastical creatures from classic fairy tales come to live in this magical piece that will get you in the mood for Halloween:
Mitch Tulloch is Senior Editor of WServerNews and is a widely recognized expert on Windows administration, deployment and virtualization. Mitch was lead author of the bestselling Windows 7 Resource Kit from Microsoft Press and has published hundreds of articles for IT pros. Mitch is also a seven-time recipient of Microsoft's Most Valuable Professional (MVP) award for his outstanding contributions in support of the global IT pro community. Mitch owns and runs an information technology content development business based in Winnipeg, Canada. For more information see www.mtit.com
Ingrid Tulloch is Associate Editor of WServerNews and was co-author of the Microsoft Encyclopedia of Networking from Microsoft Press. Ingrid is also Head of Research for our content development business and has co-developed university-level courses in Information Security Management for a Masters of Business Administration program.