Before we go into what “Ready” really means. Every great journey has a story behind it. This will be a multi-part series starting with how I joined Nutanix and evolved myself to build a world-class program called “Nutanix Ready”. Stay Tuned, Part 1 coming very soon! Rob
To continue on my last blog post on Exchange...
As I mentioned previously, I support SE’s from all over the world. And again today, I got asked what are the best practices for running Exchange on Nutanix. Funny enough, this question comes in quite often. Well, I am going to help resolve that. There’s a lot of great info out there, especially from my friend Josh Odgers, which has been leading the charge on this for a long time. Some of his posts can be controversial, but truth is always there. He’s getting a point across.
This blog post will be updated on a regular basis as things change. It will also be moved to a permanent part of the netwatch.me resources section. This is meant to be a general best practice guide to help with planning and maintaining a healthy Exchange environment on Nutanix. I will specify hypervisor specifics when required. Now on the post…..
Let’s start out with the basics…
MS Exchange on Nutanix Support
Nutanix provides a 100% supported solution for MS Exchange running on vSphere, Hyper-V or Acropolis Hypervisor using iSCSI (Block storage)
Here is a breakdown of supported configurations by hypervisor:
|vSphere (ESXi)||Use In-Guest iSCSI (Volume Groups) for full support|
|Hyper-V||Use SMB 3.0|
|AHV||Use native vDisks (iSCSI) – SVVP Certification for AHV|
Also, check out Josh’s post “Fight the FUD – Support for MS Exchange on Nutanix” that outlines this very topic. In summary, the customer has the choice to deploy in multiple configurations to suit their needs. But, one of the most often questions I get is, “does your SVVP Certification cover running Exchange on all your supported hypervisors?” The answer is not simple. The SVVP was submitted for the Acropolis Hypervisor, while this does not cover all of them, we technically are supported for all hypervisors as per Microsoft supported storage architectures. Microsoft does not specifically mention Hyperconverged, it only mentions ISCSI in regards to SAN. IMO, that covers ESXi and AHV.
Now let me explain….SAN’s are one of the biggest modern datacenter bottlenecks. Data has gravity, so co-locating storage and compute eliminates network bottlenecks = Hyperconverged is way better than SAN and hence SUPPORTED IMO 😉
To end this topic and move on, a Nutanix customer has the choice to deploy in multiple configurations to suit their needs. Being pushed to one particular hypervisor for a customer is not always in their best interest. Having choices now and later is a much better approach with the overall goal of simplifying the datacenter. As Josh said in one of his blog posts ,”Running a standard platform and storage protocol for all workloads is a simple model which reduces the unnecessary complexity of multiple protocols and/or in-guest storage configurations”, I can’t agree more with that statement. 🙂
Exchange Performance on Nutanix
Now this subject will always be controversial and potentially subject to criticism. Internal testing performed by the Nutanix Performace and Engineering team shows that AHV and Hyper-V performance are roughly the same from a hypervisor perspective and ESXi was 10% higher. That being said, usually, the next question is how is performance versus traditional SAN/NAS. And again, I have to point out, it’s all about Data Locality. Can’t change the laws of physics. Data has gravity, hence we will always beat traditional SAN architecture.
Check out Josh’s posts on “Peak Performance vs Real World – Exchange on Nutanix Acropolis Hypervisor”. It gives you a better understanding of are realistic benchmarks of Exchange in general and on Nutanix. I wholeheartedly agree with Josh when he says “Benchmarks are of little value without context specific to customer requirements!” Spending close to over 15 years building and maintain Exchange systems, I learned one hard fact, no generic simulator (like JetStress) can show real world metrics.
Data Reduction Technologies with Exchange on Nutanix
1 vDisk per Database, 1 vDisk per DB Logs
1 Container with RF2, In-Line Compression & EC-X for Databases
1 Container with RF2 for Logs
Do not use Dedupe with MS Exchange!
Microsoft does not support Data deduplication (Note: Underlying storage deduplication such as Nutanix dedupe is not mentioned, but implied)
Data Reduction Estimates:
Rule of thumb: Always size without data reduction if possible.
Conservative assumption for compression for Exchange = 1.3:1
Aggressive assumption for compression for Exchange = 1.6:1
Conservative assumption for EC-X for Exchange = 1.1:1
Aggressive assumption for EC-X for Exchange = 1.25:1
Questions to ask yourself when planning an Exchange Environment:
How many Users? e.g.: 10000, 10000, etc.
How many user profiles do you need? e.g.: 2 , Standard and Executives
How large Mailbox (excluding archiving) per User? e.g.: 1GB, 2GB , 5GB
How many messages per day do you want to support per user? Light = 50 , Medium = 100 , Heavy = 150+
Do you require site resiliency?
These are among some of the basic questions you need to answer. This is where the Exchange Server Role Calculator comes in. It’s a great tool, but like any tool, you do need to give it good input to get out good output. The function of the tool is as the name implies.
Exchange Server Role Calculator Defined
Now, at the time of this writing, version 7.8 is the latest and greatest. Now, do note, I would not call this tool perfect, but its gets you pretty close. Like anything else, the Exchange team is still learning real world behavior and this is where a good experienced Exchange engineer comes into play.
IMO..there is an Art and Science to sizing Exchange. The days of Exchange just being a simple mail server are far over. These days, it’s much more complex with supporting multiple forms of ingress and egress traffic for different functions (Mobile, Web, SMTP, Skype Integration, etc.). Each of these different functions has varying load considerations and supports more visible features like Outlook Web Access and Exchange Activesync. Also, I still am of the opinion that it does not take into consideration the number of devices that 1 mailbox services.
Considering this complexity, you can see that undersizing or oversizing can happen easily. If you size correctly at the beginning with Nutanix, then it just an easy scale out, buy as you need it situation. Then you know what happens, finally for the first time, predictability in your budgets. I remember the days, not that long ago, when I had to have a client retire a SAN, not for space constraints, but for IO constraints. And at the time, all I got from the client was “can’t we use it for something else” and ya, I’ve replied with “use it as a WSUS repository for patching the Exchange environment” 😉
During my next post, I will dive into the Exchange Role Calculator much more and go over some examples of sizing on Exchange. We’ll mainly focus on mailbox storage and then move on to other role sizing considerations. I also plan to cover the other aspects to maintain a healthy Exchange environment (i.e. Message Hygiene, Global and Local Load balancing, Integrations and End User Experience) in subsequent posts.
Below are the Office Best Practices Guides from Nutanix and some public case studies.
Until next time, Rob…..
Nutanix Offical Best Practice Guides
MS Exchange on Nutanix / vSphere Best practice guide: http://go.nutanix.com/VirtualizingMicrosoftExchangeonWeb-ScaleConvergedInfrastructure.html
Public Case Studies for Nutanix customers using Exchange
Nutanix App for Splunk
A Video Walkthrough on installation, configuration and demo of the Nutanix App for Splunk. Also, included is demo of Splunk Mobile running the Nutanix App versys Safari running Prism. To learn more about Splunk, and details on this app, check out Andre’s Leibovici @andreleibovici blog post. Happy Splunking 🙂
Until next time, Rob…
On February 16, 2016, Nutanix announced the Acropolis NOS 4.6 release and last week was available for download. Along with many enhancements, I wanted to highlight several items, including some tech preview features.
Also, checkout this excellent video with Nutanix’s Tim Isaacs and Raghu Nandan in which they go into more detail on the updates included in Acropolis 4.6 and the interviewer is my buddy Chris Brown.
Tim Isaacs and Raghu Nandan from Nutanix HQ about some of the important updates in Acropolis 4.6.
1-Click Upgrades – BIOS and BMC Firmware
The 1-Click upgrade for BIOS and BMC firmware feature is available for Acropolis hypervisor (AHV) and ESXi hypervisor host environments running on NX-xxxx G4 (Haswell) platforms only.
Acropolis App Mobility Fabric: Windows or Linux Guest Customization
Customize or clone Windows or Linux guest VMs hosted by AHV. Includes automated OS installation and custom ISOs by using sysprep (Windows) or cloudinit (Linux).
Acropolis Drivers for OpenStack
These drivers facilitate consuming the Nutanix Acropolis infrastructure as a cloud service or for use in a data center. For example, an OpenStack implementation might require using features such as single sign-on, orchestration, role-based access control, and so on. Drivers include Acropolis compute, image, volume, and network drivers.
Convert Cluster Redundancy Factor from RF-2 to RF-3
Convert a cluster created with redundancy factor 2 (RF-2) to RF-3 through the ncli cluster set-redundancy-state command. This increases the cluster fault tolerance.
Cross Hypervisor Disaster Recovery
Cross-hypervisor disaster recovery provides an ability to migrate the VMs from one hypervisor to another (ESXi to AHV or AHV to ESXi) by using the protection domain semantics of protecting VMs, taking snapshots, replicating the snapshots, and then recovering the VMs from the snapshots. To perform these operations, you need to install and configure NGT on all VMs.
Guest VM VLAN Trunking
AHV supports guest VM VLAN tagging, where the tag passes through a single port from the physical network to a VM. It allows the VLAN ID tags to be included in an Ethernet packet to be passed to the guest VM. Guest VM operating systems can use this feature to enable Virtual Guest Tagging (VGT) and simulate multiple virtual NICs.
More Backup and Data Recovery/Replication Features
- Snapshot and Async DR for volume groups.
- Application-consistent snapshots on AHV and ESXi by using the Nutanix native in-guest Volume Shadow Copy Service (VSS) agent for all VMs that support Microsoft’s VSS. Nutanix Guest Tools provides application-consistent snapshot support for Linux VMs by running specific pre-freeze and post-thaw scripts on VM quiesce.
- Integrated snapshot management from an AHV cluster to a CommVault solution
Nutanix Guest Tools
- Nutanix Guest Agent (NGA) service. Communicates with the Nutanix Controller VM.
- File Level Restore (FLR) CLI. Performs self-service file-level recovery from the VM snapshots.
- Nutanix VM Mobility Drivers. Facilitates distribution of drivers required for VM migration between ESXi and AHV, in-place hypervisor conversion, and cross-hypervisor disaster recovery (CH-DR) features.
- VSS requestor and hardware provider for Windows VMs. Enables application-consistent snapshots of AHV or ESXi Windows VMs.
- Application-consistent snapshot for Linux VMs. Supports application-consistent snapshots for Linux VMs by running specific scripts on VM quiesce.
Self-service restore allows a user to restore a file within a virtual machine from the Nutanix protected snapshot with minimal Nutanix administrator intervention. This feature is supported on Nutanix clusters running the ESXi and Acropolis hypervisors only.
Tech Preview Features
In-Place Hypervisor Conversion
This 1-click feature available through the Prism web console allows you to convert your cluster from using ESXi hosts to using AHV hosts. Guest VMs are converted to the hypervisor target format, and cluster network configurations are stored and then restored as part of the conversion process.
Native File Services
Provides file server capability within a Nutanix AHV cluster, as one or more network-attached VMs, to form a virtual file server.
To download the update, you can go to my.nutanix.com and go to support, downloads section or you can upgrade to 4.6 within Prism. Until next time, Rob
Happy New Year Everyone!!! I know Azure Stack is just around the corner, but I still get lots of questions around configuring WAP and portals. So to follow-up my Windows Azure Pack (WAP) series, I am going to talk about reconfiguring server names and ports as well as assigning trusted certificates to my WAP Portals.
If you missed other parts of the series, check links below:
Part 1 – Understanding Windows Azure Pack
Part 2 – Understanding Windows Azure Pack – Deployment Scenarios
Part 3 – Understanding Windows Azure Pack – How to guide with Express Edition on Nutanix – Environment Prep
Part 4 – Deploying Service Provider Framework on Nutanix
Part 5 – Understanding Windows Azure Pack – How to guide with Express Edition on Nutanix – Windows Azure Pack Install
“Often, you hear about something weird and un-supported, and feel like you have to share it”.
I often get calls and questions regarding backups and Exchange Server, and most backup technologies are not always working as required or as you would expect, but that’s off-topic.
One of the most common stories is that without a working Exchange Server backup, when you perform massive mailbox moves or no backup at all, the transaction logs will get piled and fill up the volume that they reside in. and then panic starts, “hey my databases were dismounted…” then of course the administrator realizes that the space on the log drive or volume has indeed ran out and now he needs to figure out what to delete. On Nutanix, we simply can solve this by extending the container that the logs live in, but what if you rely on snapshots for backups.
I had a customer reach out to me running Exchange 2007 with CCR ( Cluster Continuous Replication) on Nutanix. Yes, you heard me right, Exchange 2007 ;). They are planning on migrating to Exchange 2013 in the next year or so, but need to get from A to B for budgetary reasons until then. The only form of backup the customer has is to use Nutanix daily snapshots. The customer understands the painful process of restoring an individual mailboxes from snapshots and not having up to date recovery that logs provide along with the point in time database backup, but its a risk they are willing to take as opposed to having nothing. They reached out to me and asked, how do cleanup logs that are piling up. And so here’s where this post comes in…
My blog article suggests that you cannot sustain downtime or interruption for your users while battling with deleting log files or restoring your working backup solution. If you can sustain a downtime (should be around minutes or so) the easiest method will be to enable Circular Logging on your database / storage group – see more here –
The customer needs to be able to purge the committed logs so they don’t fill up their disk space. So how can you delete or purge Exchange server logs without any risk? well, in simple – you cannot, its built-in by design, because the whole idea of restoring an Exchange or for this matter any transnational database requires you to have a first – “full” backup of the database itself and all transaction logs that were generated since the date of the database creation date, or the last “successful” “full backup”.
Now here’s a nice method to “fake” a “full backup” or an on-demand transaction logs purge when you see you will be soon out of space, using the Exchange VSS writers and the diskshadow utility (available with Server 2008R22012R2) . This procedure also “proves” that a VSS backup for your Exchange Server will work normally.
Please note: This method was tested on an Exchange 2010 server with using a Nutanix block NX-3460-G4. Use this method on your risk. This is not supported by Nutanix or Microsoft.. You should perform a “Snapshot” before and right after this process is done.
How to manually purge Exchange server logs – with ease
This example will show you how to purge the logs for a database that is located on Drive D, the log files of the databases are also located in Drive D. we will “fake backup” drive D and this will trigger the logs to be purged.Note: If you have separated your log files and database file in different drives, or you want to include additional databases in the “backup” you must include the additional drives in the process, so in the example below, you will “Add volume e:” after “Add volume drive d:” and so on…
- Open Command prompt
- Launch Diskshadow
- Add volume d:
- (optional, add one line for each additional drive to include) Add volume X:
- Begin Backup
- End Backup
- At this step you should notice the following events in the application log indicating that the backup was indeed successful and logs will now be deleted.
Here’s some screenshots of the process:
The Diskshadow example screenshot.
ESE – Event ID 2005 – Starting a Full Shadow Copy Backup
MSexchangeIS – Exchange VSS Writer preparation.
ESE Event ID 224 – Logs are now purged
MSExchangeIS Event ID 9780 – Backup is now complete.
Final Note: although this example was tested against Exchange 2010, it should work just as fine with Exchange 2016/2013 & 2007
Until next time, Rob.