Vol. 33, #8 - June 10, 2013 - Issue #933
Clustering, Replication, and Disaster Recovery
- Editor's Corner
- From the Mailbag
- Clustering, Replication, and Disaster Recovery
- Tip of the Week
- Recommended for Learning
- Quote of the Week
- Admin Toolbox
- Admin Tools We Think You Shouldn't Be Without
- Events Calendar
- Webcast Calendar
- Register for Webcasts
- Tech Briefing
- Data tiering and service level agreements
- Avoiding storage failures
- Enterprise storage best practices
- Volume and File System Considerations for Windows Server 2012 Hyper-V
- The Microsoft Virtual Machine Converter (MVMC) and MAT Architecture #SCVMM #Hyperv #VMware
- VIDEO: Windows Server 2008 R2 Dcpromo "Operating System Compatibility" Message
- Product Review: MailStore Server v8.0
- Planning and migrating a small organization from Exchange 2007 to 2013 (Part 1)
- Migrating Public Folders to Exchange 2013 (Part 1)
- GFI WebMonitor for ISA/TMG Voted ISAserver.org Readers' Choice Award Winner - Access Control
- Implementing Windows Server 2012 DirectAccess behind Forefront TMG (Part 1)
- Windows Server News
- Do you need to reevaluate your BYOD policy?
- Understanding Windows Group Policy management for virtual desktops
- Five tips for building a VMware virtual infrastructure
- Windows Phone 8 security should be part of any mobile device strategy
- WServerNews FAVE Links
- This Week's Links We Like. Fun Stuff.
- WServerNews - Product of the Week
- 3 Free Tools. 1 Easy Download. Active Directory Admin Simplified.
- SAVE THIS NEWSLETTERso you can refer back to it later for helpful tips, tools and resources!
- FORWARD THIS NEWSLETTER to a colleague who you think might find it useful!
- SEND YOUR FEEDBACK to [email protected] if you have any comments or suggestions!
This week's newsletter is all about how you can use some of the new Hyper-V and clustering capabilities in Windows Server 2012 to implement a better disaster recovery plan for your organization's datacenter. We welcome Symon Perriman, Senior Technical Evangelist at Microsoft, who has provided us with a guest editorial on this topic.
Fortunately, the organization where Dilbert works also has a disaster recovery plan in place. But as Alice explains in this comic, there's something missing from their plan:
Does that ring a bell with your organization too?
From the Mailbag
We accumulated quite a bit of mail while we were away on our vacation in Brussels, and unfortunately we won't be able to publish all of it. Let's start with some reader feedback from last week's editorial Tiptoeing Towards Windows 7 (Issue #932). A reader named Joe had the following comments:
Mitch, your Figure 4 illustration on the best way to open a command prompt is not the most realistic scenario. You have to remember the Start Screen like the Start Menu is never active while you are doing anything. So the more realistic scenario is the desktop or a Windows 8 app and from that perspective the logic is the exact same i.e. hit the Windows key and type 'cmd'.
There are two scenario's where the Start Screen is up: On boot-up/wake-up, or if you are perusing the Start Screen for tile updates which means you had to explicitly hit the Windows key to get there.
I embarrassedly realized when I was switching to Windows 8 that my skill set for client interaction was a Windows 95 skill set. Had I updated my skill set when I adopted Vista or Windows 7 (meaning use the Window key) the only transition would have been to come to grips that the Start Menu was now a Start Screen.
Not to say there are not some frustrations, but after the wonderment of Windows 95 wore off there were frustrations with that OS also.
Some good points there, especially about having a Win95-centric way of thinking. Although we upgraded our computers here from Windows XP to Windows 7 back in 2010, I think I still use the Windows XP way of doing a lot of things on my desktop computer...
Another reader named Phill pointed out another good way of opening a command prompt in Windows 8:
I'm not an IT guy, but in the example you use you can open a command prompt from either environment by pressing the windows key + x, then pressing c. I have a feeling that many commands will be added to the winkey + X menu by utilities not yet written...
A reader named Art suggested a utility you can use to avoid the new Start screen entirely:
Hi Mitch, Love your newsletter! Ref you use of Win8 article, the question you fail to ask is, what advantage do you derive from even bothering with the new Win8 desktop? I use Start8 and have it set up to go right to the old desktop. I've used the new desktop all of once in almost six months. There may be some value there, but not enough to motivate me to learn how to use it. If they improve the apps enough in 8.1 perhaps I'll have another go at it. So I get all the advantages of multi-display integration, faster reboot, Flash disk integration, etc. with no pain.
Here's a link to where you can get Start8:
Another reader named Kevin who is an IT and operations manager in Australia also recommended Start8 as follows:
As the Operations & IT Manager for a 40+ user family business in Australia, I need to be very discerning when it comes to investing company funds on IT enhancements. Like most other businesses, you try to resist "dipping your toes" on a new OS until it has been fully shaken down by larger organizations who (presumably) can afford possible inconveniences a new OS may entail.
The functionality parallels between the iPod Touch OS & using Windows 8 seem to support the thought that Windows 8 is more aligned to tablet, rather than desktop, use. I too would hope that the next release, or perhaps service pack, will return some of the more familiar aspects of Windows 7 which business users have continued to use.
As I'm sure you would agree, for a business, and in particular for small businesses, an increase in productivity is generally a key factor when justifying IT expenses. From a productivity perspective, I'd far prefer to click on an 'X' in the top right hand corner than moving the cursor to the top of the screen, hesitate and then slide from top to bottom, simply to close an application.
Above all though, when there is such a radical OS change, as has occurred with Windows 8, it is the learning curve and / or re-training for users which smaller businesses cringe at. Indeed, for an organization like ours, the re-education of our users would likely be more costly than the OS itself.
Since we maintain a structured hardware succession plan, we have already migrated a couple of users over to Windows 8. The new hardware came with Windows 8 pre-installed, so it made sense to "bite the bullet". A worthwhile point here is that following on from one of your earlier articles, we also installed the Start 8 program. The cost was $4.99 and it took pretty much all of the pain out of migrating users to Windows 8, as the majority of the system looks & feels just like Windows 7. Yes, there are odd times where the system will jump back to Windows 8, however if the users click in the lower left hand corner, they return immediately to the Windows 7 environment. This process enjoys all of the security & performance features built into Windows 8, while providing the users with the familiar comfort of Windows 7. This in turn significantly reduced the upgrade training. I would highly recommend this path for other small businesses who can't avoid migrating a system to Windows 8.
I thought other readers may appreciate the perspective of a small family run business where expenditure must be rigorously scrutinized. Keep up the great work in keeping us all informed of the ever changing world of IT.
A reader named Tom shared some additional complaints about the new "modern" desktop in Windows 8:
I've only had my Windows 8 system since November, and it is my secondary system since it is a laptop that I used on vacations and for my non-work meetings, but I find that I have to really MEMORIZE everything with the "Modern" desktop.
There really is no organization that you can easily do. You can group things on the modern, but you still have to MEMORIZE where you put things in what groups - and you cannot easily SEE all the groups on the interface at all - you HAVE to scroll to find anything. Which I find very UN-productive.
You mentioned that you can just type the name of the app to find it - but what name is it really registered under? When you install any app does it TELL you what name it is going to use so you can find it? No - you have to GUESS what name it is using and hope that what YOU think it is called is what the developers called it when they put that info into the registry. If you do NOT know the name of the app how in the world can you find it? Oh yes, call up ALL apps and then you have to read EACH AND EVERY ONE OF THEM in order to hope you find the one you want - AND - that they did not install it under the name of the company which often is NOT the name of the application!
I agree that some people can NEVER organize anything in real life or in a computer - and for them just typing it in and hoping to find it may work - but I would suspect that the vast majority of people DO categorize and organize and forcing them to just type in some random word in hopes that the programmers WHO DO CATEGORIZE AND ORGANIZE thoughts put the name of the app into the system the same way the end user thinks of the app name is just plain stupid.
Also by NOT allowing people to categorize and put things into the system the way they think will cause ALL sorts of productivity issues later on when they try and find it AGAIN.
I have 3 TB worth of data, around 30,000 files counting my photos, and just searching for a word to find that specific file would never work - SO many files would statistically have that word in the meta data that you get hundreds of hits - and then you have to read EACH AND EVERY ONE OF THEM to find the correct file that you want.
The "usability experts" at Apple NOR Microsoft every mention this problem in their ads or docs when saying how great "just type it in" works.
In a business environment telling people they can ONLY look at one doc at a time - and then task screen switch to the others and then switch back to the one they are really working on instead of seeing multiple up at the same time is going to KILL productivity - you will spend more time switching back and forth than doing real work. Right now I have your email up, a news article, Outlook, Citrix desktop and I can see all of them on my screen - yet in Windows 8 I would have at most 2 (right now) and one of them would be 1/3 screen and be so unusable to be worthless. And MS thinks this is a better productive enhancement? If it was a PHONE or just a CONSUMER of data at a single point in time to take NO action on okay, but to create and perform REAL work in a business environment - laughable.
I can relate to some of what Tom says above. Fortunately it looks like some of the changes coming in Windows 8.1 will be addressing some of the usability issues in Windows 8. In fact, the day after I posted last week's newsletter into our publishing system, the following must-read post appeared on Microsoft's Blogging Windows website:
Mary Jo Foley also posted some interesting news items last week about Windows 8.1 on her ZDNet blog:
Her first item especially intrigued me because having the same background on the desktop and Start screen will probably make the transition between them much less jarring. In fact, she actually used the word "jarring" in her article to describe how it feels moving between these two environments, so I guess I'm not the only one that feels this way.
A reader named Neil from the UK had some advice for me about not giving up too soon on Windows 8:
Mitch, I never email people about articles I read but I've just read WServerNews this week and felt compelled to break my mold!
I've been using Windows 8 for several months now and I experienced a reverse J curve of annoyance. In the first week I could've thrown my laptop out of the window several times a day, the second week less so and so on. I read articles about how it worked, what keys to use, the difference between native and desktop apps and I grew to love it. I am now in the position where I wouldn't let Windows 8 go if I was paid to do so, so much so that I am replacing my laptops (I have Windows and Mac) & iPad with a Surface Pro and my iPhone 4S with a Nokia Lumia 925 (when it comes out) -- 4 devices down to 2 (and an external hard drive).
I am worried about the constant whining about the lack of a start button. It's like people don't want to let go of their comfort blanket, they always have to have something to moan about or they just reject change. These people are IT people that should embrace change and progression! I have 10 year old twin daughters who absolutely love Windows 8, they took to it instantly from Windows 7 -- they said they find Windows 8 easier and clearer because the apps use the whole screen and have no interference from other windows. They learnt about the various Windows Key combinations by reading a Windows 8 app and I now watch them navigate around like professionals. Similar, my parents, both retired for some time, took to it equally well -- again, after doing some background reading.
My key point here is that you shouldn't judge Windows 8 by an ill-educated stab in the dark at it, it's not a radical move but it is different. You should learn about it, get to know it and remember that no pain = no gain!
Oh, to be young again...
Brett from Australia had a tip for me about taking screenshots of my iPod (see Figure 3 in the last issue):
Just a tip with your photos of the screens of the iPhone search -- if you press the home button and the power button at the same time, the screen will be saved as a picture in your camera roll. Much easier than using a camera.
A reader named Michael who is a Microsoft guy and an MCSE shared the following thoughtful comment:
Mitch, regarding your comments on Windows 8.
In a nutshell, I think you just got it.
I think that Microsoft was attempting a paradigm shift with Windows 8.
They just didn't do a very good job explaining the real reason you should switch to Windows 8.
Being a Microsoft person, I understand that change is good but I also understand that the world is a different place than it was when Windows 95 came out. I like to think that was the original paradigm shift and along with 32-bit preemptive multitasking like OS/2 it offered users 'plug and play.'
This allowed users to not have to spend so much time in DOS configuring hardware IRQ's.
I think the changes of Windows 8 are in a similar place for Microsoft but an important piece was forgotten. They didn't explain what the benefits of upgrading were that made sense to your average user. That when folks changed from DOS or Windows 3.1 that the regular user would be able to perform tasks that used to require hardware geeks.
A reader named Carl gives us another reason (Client Hyper-V) why migrating to Windows 8 can bring valuable benefits:
I realize "your mileage may very" applies here, but I am really quite perplexed at all the wailing and gnashing of teeth going on about Windows 8 right now. I've been using it since it was released. I adapted to the new start screen after about a week and have not looked back. I upgraded to get Hyper-V. I now can run 6 VMs for software testing with no significant performance hit (except when booting.)
In regards to the Start screen, I actually have no problem with switching to it because of the hot corners and the fact that the tiles are bigger "click targets" which means I can start a program from the Start screen faster than I ever could using the Start Menu in Window 7.
Are there issues with Windows 8? Absolutely, but no more than any other 1.0 release. The All Apps view kind of sucks, as does the Microsoft Store. My biggest concern with the new Metro apps is that I have difficulty visualizing how complex productivity programs such as AutoCAD or Corel DRAW would be any better, or even possible using the new UI. But, I'll give them the benefit of doubt for now because right now Windows 8 provides a better desktop OS than Windows 7.
Many other readers sent us feedback concerning the last issue and we're sorry that we can't include you all in this issue of the Mailbag.
Moving on then...
In the issue RAID Controllers (Issue #928) we included a guest editorial by Mike Pepe who works as a Service Engineer for the Bing Information Platform at Microsoft. A couple of readers asked for some clarifications, so here are their questions along with Mike's replies.
A reader named Bob said:
Excellent basic explanation of RAID technology. I do have one question, though.
When Mike calculates his error rate prediction for the array of 4TB disks, it seems he is equating an error rate of 1 unrecoverable bit in every 10E14 READS (emphasis mine) to the total number of bits in 16TB, which is 16,000,000,000,000*8bits, or 1.28x10E14 bits. Depending on the block size of the disk in question, I believe we read the data in chunks of around 512 byte sectors at a time, which reduces the number of reads needed to process 16TB by a factor of 4096. In addition, in an enterprise server environment, I would expect the drives to be SAS, which has failure rates more like 1x10e16, as opposed to SATA at 1x10e14.
Curious to see Mike's thoughts on this.
Mike replied as follows:
I'm not surprised to see this sort of question as reliability data can be tricky to comprehend.
The data we used came from a drive manufacturer's data that was sent to us while we were evaluating engineering samples of their latest drive. Unfortunately I can't share the data but I can tell you that their reliability metric is defined as "less than 10 uncorrectable read errors in 10^16 bits read." Since this is preproduction data we used the older metric from a similar, earlier enterprise SAS drive series which had error rates defined between one in 10^14 and 10^15 bits read.
The drive manufacturer's metrics are defined as errors per bits read, not in per sectors read or read operations. We're using the numbers on the lower end of the scale, as we feel a good engineer should to better protect against the worst-case-scenario.
A 4TB drive has 32,000,000,000,000 (3.2x10^13) bits on it. 1x10^14 bits equates to 12.5 TB, and a reliability number ten times greater (1 in 10^15) represents 125TB. In the example, yes, a 16TB array has 1.28x10^14 bits. Using even the higher end numbers where there is one uncorrectable read error in 10^15 bits means that on average you'll have one uncorrectable read error every 8 times the 16TB array is completely read, which is still something one should design for.
Another reader named Simon said:
Fascinating article! I have been teaching RAID for years and this confirmed some of what I already knew and extended my knowledge as well.
Two questions though:
- I have searched high and low for a definitive article on what "Triple Mirroring" is. Microsoft says it requires 5 drives and can tolerate 2 failures but I cannot find anywhere how it works.
- Are you sure about the failure rate for RAID 0? A quick look at Wikipedia (which is of course entirely trustworthy) gives a more complex calculation.
Thanks for all the work and updates. I shall be signing up for the Azure JumpStart - thanks for the heads up.
I did talk to my spaces expert and unfortunately we couldn't come up with a good document that expresses how a three way copy actually works. The responder is correct though, in order to do three-way copying you do need to have at least 5 disks. I'm not totally an expert in how spaces works, but the reason has to do with maintaining a majority in the event of a two-drive failure. The way spaces would write the data out across 5 drives in a 3-copy scenario means that two drives can be lost with no loss of data.
As far as the reliability statistics go, yes, the responder is also correct. When dealing with statistics and probability more complex calculations are really necessary to come up with truly statistical accurate numbers, but given the scope of the audience (and the fact I'm not a statistics whiz) we kept it simple and treated each drive as an individual and not the "true" reliability as calculated for the whole system. The Wikipedia article is actually a really good reference on the subject, but we took the simplified "Additively" approach since each drive presumably has the same specified rate of failure.
Finally, several readers responded negatively with what might be characterized as "rants" concerning Jeffrey Hick's guest editorial in the issue Say No to PowerShell in the Enterprise (Issue #929). But rather than include these readers' comments, we'll just post Jeffrey's response from his reading them:
I have to say I am amazed by the response to my recent editorial, "Say No to PowerShell in the Enterprise". Many of you appreciated my subtle and maybe even subversive approach. I have to admit I like playing devil's advocate and stirring things up. That's when the most interesting things happen. But more than a few of you took my comments much too literally. First off, for those of you who don't know me, I am a long-time PowerShell MVP, the secretary of PowerShell.org and make my living from PowerShell so I don't think you'll find a more ardent proponent. I didn't mean for some of the points to come off as glib or trite. I was trying to enumerate a short list of reasons that actually are *why* PowerShell is so important in a short amount of space. The bottom line is that you have to manage a Windows environment and want to be as efficient as possible, then you need to be learning and using PowerShell. If you would like to continue this conversation, I recommend the forum at PowerShell.org:
However, one reader did send us the following witty comment concerning Jeffrey's article:
All I'm going to say is, if I wanted to manage things with a CLI...I'd be on Unix. Which, on many flavors, you can also manage graphically.
And all I can say to that is, thanks for the feedback!
Clustering, Replication, and Disaster Recovery
And now on to our guest editorial by Symon Perriman...
Multi-Site clustering places Failover Clustering nodes at different physical locations to maintain high-availability in the event of an entire datacenter becoming unavailable due to a disaster, power outage, attack or some other unexpected event. In this situation, nodes at a second site will detect that the primary site has failed and bring the resource online, either manually or automatically. Spanning a Failover Clustering across multiple locations for disaster recovery is also known as "stretched clustering" or "geographically dispersed clustering". A variety of security, networking, storage and high-availability goals need to be considered when designing, deploying, managing and testing for disaster recovery, and so there is no single architecture that meets everyone's disaster recovery needs.
With the introduction of the Hyper-V Replica in Windows Server 2012, Microsoft has provided a free in-box replication solution for Hyper-V Virtual Machines (VMs). While most traditional multi-site clustering principles are still used with the Hyper-V Replica, this technology introduces a new architectural design for multi-site clustering. First, let's review the traditional multi-site design which has been supported since Windows Server 2003.
Traditional Multi-Site Clustering
There are numerous considerations for multi-site clusters, including servers, security, networking and storage. In the following diagram you can see a traditional multi-site deployment architecture with a single cluster which spans both sites. It is also important that high-availability can still be maintained at every location if after a disaster has already partitioned the datacenter, so each site must have and use traditional clustering best practices for servers (cluster validation, reserved capacity to support all roles after failover, etc.), networking (NIC teaming, redundant networks, etc.) and storage (MPIO, RAID, etc.). To learn about all the multi-site design considerations visit here:
Figure 1: Traditional multi-site clustering.
Multi-site clusters can deployed on different subnets on each site, or on a single subnet by stretching a VLAN across the sites. Both topologies are fully supported and work with Hyper-V, live migration and Cluster Shared Volumes. While configuring a VLAN adds some management overhead, it minimizes client downtime during a failover, so it is generally recommended to provide a higher level of availability. The time that it takes for the system to recover and for clients to reconnect to the workload is called the Recovery Time Objective (RTO). Several factors impact the RTO, such as the time it takes to bring the VM online after a failover, DNS replication and client reconnection speed.
It is important to understand how DNS works in a multi-site cluster to minimize downtime to clients after a cross-site failover. When a client or application wants to connect to a role running on the cluster, the user request an A/AAAA record from the DNS server which includes the IP Address of the service or VM. By default, this DNS record expires every 20 minutes, so the client will pull a new record with any updated information three times each hour. When a cross-site failover happens on the same subnet (by using a VLAN) the workload will keep the same IP Address, so the client should be able to immediately reconnect as the information in their local DNS record is still accurate. However when a failover happens and the workload moves to a different subnet, then that workload will have a different IP Address, the local DNS record will still have the old IP Address, and the client will not be able to reconnect until they get the new IP Address. This happens once the stale record expires and they receive a new DNS record with the updated IP Address, however slow DNS replication can further delay reconnection times as DNS servers have to also update each other with the new record. However there are several techniques used to reduce the time it takes for client to reconnect after a failover.
- Configure local failover first: It is recommended to have at least two nodes in every site and to configure the workloads to stay on the same subnet at the local site first before failing over to the secondary site. In a multi-subnet cluster this enables the client to remain connected using the same IP Address. Local failover can be configured using the preferred owner and possible owner properties for each cluster group. All the nodes at the primary site would be set as preferred owners and the nodes at the secondary site as possible owners.
- Reduce HostRecordTTL time: This cluster group property will adjust the time which a DNS record remains on a client's system before a new record is pulled down from the DNS server. By increasing the frequency that the DNS record is updated, the client will be able to see and access the new IP address sooner, allowing them to reconnect to the workload faster. By default, the record gets updated every 20 minutes, however in a multi-subnet cluster it is recommended to lower this value so that the record is replaced more frequently. For example, Exchange Server recommends a HostRecordTTL value of 5 minutes.
- Enable the RegisterAllProvidersIP setting: This cluster group property will adjust the information stored in DNS records for clustered workloads by including multiple additional IP Addresses which the workload can use (one for each subnet). Assuming that the client has the logic to try to connect to additional IP Addresses if the first address is unavailable, then this enables the client to reconnect much quicker in a cross-subnet failover. After a failover the client will first try to connect to the original IP Address they were using, and then they will try to connect to other IP Addresses in the DNS record and they should be successful.
Security & Reliability Considerations
Since network traffic is likely traveling outside the datacenter and across the "public" Internet, security planning is important for both the data and the cross-site cluster communication. Securing the application data that is copied between the sites is generally handled by the storage replication mechanism. Securing the cluster communication is also important for resiliency to ensure that cluster communication is not compromised by malicious attempts to spoof messages between cluster nodes. By default, communication between nodes is signed which means that the sender/= and recipient pair are authenticated. However all cluster communication can be fully encrypted or disabled entirely, and there may be a slight change in performance depending on which of the three security levels is selected. This option can be configured by the cluster property SecurityLevel.
When stretching cluster nodes across any distance network latency can play an important role. This is because nodes are continually monitoring the health of each other using pings called "heartbeats". The cluster will consider a node (or site) unavailable if the node(s) does not respond to several consecutive heartbeats. Even if all the nodes are healthy, if too many of these heartbeats are missed due to network latency, some nodes could think that other nodes are unavailable and trigger a false failover. When using multiple subnets within a cluster the frequency (how often heartbeats are sent) and resiliency (how many heartbeats can be missed) of these heartbeats is fully configurable. These settings can be adjusted using the cluster properties SameSubnetDelay, SameSubnetThreshold, CrossSubnetDelay, and CrossSubnetThreshold.
In a multi-site cluster the data must always be replicated between the storage arrays in both locations so that the application's data is always available, regardless of which site is hosting the workload. There are multiple options for storage and what you use depends on your cost, distance, and data loss goals. Since the method for quorum witness disk replication (or lack thereof) can vary from vendor, it is best to use their recommendation for the quorum configuration.
The two logical types of replication can happen at a block-level or file-level. Block-level replication will happen at the hardware and storage level by copying logical blocks of data between sites. This replication solution is provided by all major storage vendors which support Failover Clustering. File-based replication will happen at the software level by copying files between sites. This replication solution is provided by several software companies which support Failover Clustering.
To keep the data consistent across both sites, the replication frequency is important. This affect the amount of data that could be lost during a disaster and is known as the Recovery Point Objective (RPO). For example, if the data is replicated between sites every 15 minutes, but a crash happened 10 minutes since the data was last replicated, then it is possible to lose up to 10 minutes of data. This type of replication is called asynchronous and the writes to the secondary storage device happen in the background, but are not guaranteed. Asynchronous replication solutions are generally cheaper, faster and can support sites that are stretched across greater distances, however it is possible to have several minutes of data loss. Alternatively synchronous replication will write changes to both the primary and the secondary location, before the write transaction is committed. This ensures that there is no data loss and the information is always consistent at both sites, however this solution often comes with a performance overhead and the maximum distance between sites is usually shorter.
If you are using traditional multi-site clustering for Hyper-V, then replication between the Cluster Shared Volumes (CSV) disks must also be considered. While most replication vendors support CSV, it is important to confirm this to ensure that their replication driver is compatible with CSV.
While most clustered roles use this traditional multi-site cluster architecture, it is important to note that some roles have unique deployment considerations due to their data replication mechanism. For example, SQL Server has several built-in database replication technologies such as mirroring; Exchange Server uses log shipping; and DFS-Replication uses its own built-in method. DFS-R is actually not supported as a replication solution in a multi-site cluster as this technology only replicates data when the file is closed. This means that for workloads such as SQL and Hyper-V where the data file remains open, the data would never be replicated to the other side and hence unavailable after the failover.
Multi-Site Clustering with the Hyper-V Replica
While traditional multi-site clustering can still be used for Hyper-V in Windows Server 2012, the most recent server release introduced the Hyper-V Replica feature which allows administrators to replicate their Hyper-V VMs to different servers or sites. This in-box solution will work with any supported storage or guest OS workload by copying the VHD/X file and any subsequent changes between any combination of standalone hosts or clusters running Windows Server 2012 Hyper-V or Microsoft Hyper-V Server 2012.
However the Hyper-V Replica's replication engine behaves differently when it runs on a cluster as we make the replication service itself highly-available which is known as the Hyper-V Replica Broker (HVRB). The HVRB is created and managed on a cluster like any other clustered role, which means there is a single instance that is highly-available and fails over between the different cluster nodes. This single instance of the HVRB means that it is not possible to use the Hyper-V Replica to copy VMs between nodes within a single cluster. Hyper-V replication can happen between two standalone hosts, between the HVRB in a cluster and a standalone host, or between two HVRBs in two clusters. This changes the configuration for a multi-site Hyper-V cluster from a single cluster than spans both sites to separate clusters in each site. Once the Hyper-V Replica Brokers have been deployed at both sites, replication can then be configured the same way as if using a standalone Hyper-V Replica. Storage is still required at both sites, but no 3rd party replication solution is needed.
Figure 2: Multi-site clustering with the Hyper-V Replica
Finally, it is essential to test the solution before it is placed into production. Hyper-V Replica has multiple built-in test tools to allow the verification of both replication and reverse replication. By combing the Hyper-V Replica with multi-site Failover Clustering a datacenter can now easily be configured for disaster recovery using a free in-box solution from Windows Server 2012. For additional information about the Hyper-V Replica visit here:
or check out Microsoft's free guided training from the Microsoft Virtual Academy here:
About Symon Perriman
Symon Perriman is a Senior Technical Evangelist at Microsoft Corporation. You can follow him on Twitter here:
Send us feedback
Got feedback about anything in this issue? Let us know at [email protected]
Tip of the Week
GOT A TIP you'd like to share with our readers? Email us at [email protected]
A reader named John sent us some feedback concerning last week's issue, and we thought his comments would make a terrific tip. Here's any easy way to give your mouse a break when launching apps on Windows 7:
I don't like to mouse around through menus. I have my favorite apps and that is about all I do. Office type work.
I assign keyboard shortcuts directly to unique apps, and I do that for every computer I use. Makes no difference whether I am using a 10 year old laptop or my newest Gateway geewhiz 8 gigabyte RAM speed demon.
- Control-Alt-G Google Search, also Control-Control does the same thing. Great app.
- Control-Alt-W starts Word
- Control-Alt-X starts Excel
- Control-Alt-T starts Thunderbird email
- Control-Alt-I starts Internet Explorer(x)
- Control-Alt-F starts Firefox
- Control-Alt-E start Explorer(2) a multi-pane version of Windows. Explorer based on the old one in Win3, Win5, etc.... it is great.
- WindowsKey R brings up the Command Prompt (who needs a mouse?)
- WindowsKey E brings up Windows Explorer
- WindowsKey T displays the Task Bar
Just run thru the keyboard with the Windows key held down and look at the stuff that has been there since day 1.
Recommended for Learning
This week we have a few business titles from O'Reilly that you might want to check out if you are a business owner as well as an IT nerd.
Lean Analytics: Use Data to Build a Better Startup Faster
I'm a big believer in measuring things. If something isn't measurable, it's probably not real. I track my productivity to keep my business successful and my fitness/weightloss progress to ensure I'm staying health. This book might help you face reality by finding metrics you can use to grow your business. It's got good reviews on Amazon, check it out using the link above.
Shipping Greatness: Practical lessons on building and launching outstanding software, learned on the job at Google and Amazon
Tons of tips distilled from the real life experience of the author. Takes you through the whole process of shipping software step by step. Can also apply to other areas of managing projects.
Loyalty 3.0: How to Revolutionize Customer and Employee Engagement with Big data and Gamification
Wikipedia says that "gamification" is the use of game thinking and game mechanics in a non-game context in order to engage users and solve problems. We all love games, some people even say that life itself is a game. What does this mean for success in business? The author shows how you can use gamification techniques to build and maintain customer loyalty to drive profits and growth for your business. I had a chance to review a preview copy of this title and I found the case studies especially interesting. Check it out.
We also have the following announcement from the Microsoft Virtual Academy:
Learn about Windows 8 Security from Microsoft Premier Field Engineers
This free on-demand course from Microsoft Virtual Academy (MVA) addresses key insights, capabilities, and best practices for managing & securing Windows 8 clients. Learn about Windows 8 security improvements over Windows XP, Windows Vista, and Windows 7.
Quote of the Week
"If you write something truthful about your own experience then it has some universal appeal." - Julie Delpy from an interview in Hello! magazine about her films with Ethan Hawke
Until next week,
Note to subscribers: If for some reason you don’t receive your weekly issue of this newsletter, please notify us at [email protected] and we’ll try to troubleshoot things from our end.
Admin Tools We Think You Shouldn't Be Without
SolarWinds free NTFS Permissions Analyzer for Active Directory - get a complete hierarchical view of the effective permissions & access rights for a specific file folder or share drive. Download today!
Free Tool: Idera Server Backup Free – fast, disk-based continuous data protection for Windows and Linux servers – backs up and restore files in seconds
Download 2X ApplicationServer XG to deliver virtual desktops and applications from a central location, providing continuous availability, resource-based load balancing and complete end-to-end network transparency for administrators.
Exclaimer Mail Archiver is a fully featured, competitively priced archiving solution for Exchange that's easy to set up and maintain. It uses file system-based storage so it doesn't require SQL.
Plugable USB3-SATA-UASP1 USB 3.0/2.0 SATA Lay Flat Docking Station:
Microsoft Build on June 26-28, 2013 in San Francisco, USA
Microsoft Worldwide Partner Conference on July 7-11, 2013 in Houston, USA
Microsoft TechEd Europe on June 25-28, 2013 in Madrid, Spain
NEW! Microsoft TechEd Australia on September 3-6, 2013 in Gold Coast, Australia
NEW! Microsoft TechEd New Zealand on September 10-13, 2013 in Auckland, New Zealand
Add your event
PLANNING A CONFERENCE OR OTHER EVENT you'd like to tell our 100,000 subscribers about? Contact [email protected]
Register for Webcasts
Add your Webcast
PLANNING A WEBCAST you'd like to tell our 100,000 subscribers about? Contact [email protected]
Data tiering and service level agreements (WindowsNetworking.com)
Mitch Tulloch looks at how implementing SLAs and bill-back can help drive a data tiering solution towards success.
Avoiding storage failures (WindowsNetworking.com)
Mitch Tulloch describes some of the common forms of storage failures that happen in enterprise IT environments, how to avoid them, and how to deal with them when they occur.
Enterprise storage best practices (WindowsNetworking.com)
Mitch Tulloch summarizes some best practices you can implement that can increase storage performance, availability, searchability, and reliability in your IT environment.
Volume and File System Considerations for Windows Server 2012 Hyper-V (VirtualizationAdmin.com)
Brien M. Posey lists some of the more noteworthy storage improvements and general considerations that should be taken into account when implementing Windows Server 2012 and/or Hyper-V.
The Microsoft Virtual Machine Converter (MVMC) and MAT Architecture #SCVMM #Hyperv #VMware (James van den Berg, MVP/IT Pro)
James wrote a blog how to convert VMs and physical Systems to run them under a Hyper-V environment.
VIDEO: Windows Server 2008 R2 Dcpromo "Operating System Compatibility" Message (WindowsSecurity.com)
Derek Melber explains the meaning behind Windows Server 2008 R2 Dcpromo "Operating System Compatibility" message.
Product Review: MailStore Server v8.0 (MSExchange.org)
J. Peter Bruzzese takes a look at MailStore Server v8.0.
Planning and migrating a small organization from Exchange 2007 to 2013 (Part 1) (MSExchange.org)
Steve Goodman begins a new article series that will focus on how to migrate from Exchange 2007 to Exchange 2013 including planning and configuring co-existence.
Migrating Public Folders to Exchange 2013 (Part 1) (MSExchange.org)
Nuno Mota begins a two-part article that talks about and demonstrates how to migrate Public Folders from Exchange 2010 into Exchange 2013.
GFI WebMonitor for ISA/TMG Voted ISAserver.org Readers' Choice Award Winner - Access Control (ISAserver.org)
GFI WebMonitor for ISA/TMG was selected the winner in the Access Control category of the ISAserver.org Readers' Choice Awards. Chaperon for ISA 2004/06 and Burstek WebFilter for ISA/TMG were runner-up and second runner-up respectively.
Implementing Windows Server 2012 DirectAccess behind Forefront TMG (Part 1) (ISAserver.org)
Marc Grote begins a two-part article that explains how to configure Windows Server 2012 as a DirectAccess Server. Part 1 covers the basic parts and the technology behind Windows Server 2012 DirectAccess.
Thanks to Florian Klaffenbach for providing some of the items in this section. Be sure to check out Flo's Datacenter Report:
Do you need to reevaluate your BYOD policy?
While the BYOD trend can be beneficial to organizations as it often increases employee productivity, it also introduces a number of key security pain points for IT pros. Inside, discover essential tactics that can help you develop a BYOD policy for your organization that balances the risks and rewards.
Understanding Windows Group Policy management for virtual desktops
Managing your organization’s various virtual desktops can be complex and time-consuming, but leveraging Windows Group Policy can help significantly simplify this task. Learn why Group Policy works so well with VDI deployments and discover the key advantages that this approach can offer.
Five tips for building a VMware virtual infrastructure
To build a high performance, reliable VMware virtual infrastructure, there are many factors to consider, steps to take, and pitfalls to avoid. Access this tip to explore five essential VMware virtual infrastructure planning and design tips that every VMware admin should know.
Windows Phone 8 security should be part of any mobile device strategy
For today’s organizations, having an effective mobile device strategy in place is essential, and the iPhone isn’t all you need to worry about – the Windows Phone 8 is just as risk-prone. Inside this tip, discover key factors to consider to ensure Windows Phone 8 security in your organization.
This Week's Links We Like. Tips, Hints And Fun Stuff
GOT FUN VIDEOS or other fun links to suggest you'd like to recommend? Email us at [email protected]
This dream garage belongs to billionaire Craig Jackson, and within it are some of the most valuable cars in the world:
"The Inbox" (Short Film - 3 min) What online scams would look like if they happened in the real world:
To succeed against the powerful car lobby in 1901, Henry Ford challenges the champion driver Alexander Winton to the famous "Race That Changed the World."
A Russian bear knows some impressive tricks. Such things do not exist only in fairy tales. That's life in Russia, where your friend can be a real bear:
'Attraction' Shadow Theatre stunned the judges and the audience of Britain’s Got Talent with its graceful and moving story:
WServerNews - Editors
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 Tullochis 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.