Tag Archives: VSAN

vSAN Licensing: Expanding Clusters and Setting off Alerts

vSAN licenses are assigned per-cluster, so whilst the total number of vSAN licenses in a vCenter inventory might match the total number of vSAN host CPUs, they may not be assigned correctly to the clusters. This will trigger the critical vCenter alarm “License inventory monitoring”. This will also occur if a cluster is expanded (e.g. additional hosts and vSAN licenses are purchased) as it’s only possible to assign one license key to each cluster.

In the example screenshot here we have 2 vSAN license keys, each valid for 8 CPU. The total 16 CPU capacity matches the 16 CPU in use. However, one cluster has 10 CPUs (i.e. 5 dual socket hosts) and the other only 6 CPU. Therefore one license key is 2 CPU oversubscribed whilst the other has 2 free.


The method to resolve this is to use the MyVMware license portal to split and merge your pool of vSAN licenses until you have license keys where the capacities match your cluster sizes, and then re-license the vSAN environment. This VMware KB article explains in detail how to do this divide, and merging is a similar process on the same interface: https://kb.vmware.com/s/article/2006972

In the example screenshot above, this was a case of splitting one of the 8 CPUs into a 6 and a 2, and then merging that 2 with the remaining 8 CPU key to create a 10 CPU key and a 6 CPU key which matched the cluster sizes. The old 8-CPU key that was split was removed from the license inventory.



vSAN Editions

There are three major editions* of vSAN 6.6 available – “Standard”, “Advanced”, and “Enterprise”- so other than price what’s the difference between the three?

*aside from the “vSAN for Desktop” and “vSAN for ROBO” lines which address specific use cases- look out for later posts focusing on these.


Standard is the base offering and, like it’s more expensive cousins, is licensed per socket the same way that vSphere is- so a dual-socket host needs two licenses for vSphere and two licenses for vSAN. This edition gives you all the core features of the VMware software-defined storage platform; distributing VM storage across the converged hosts, supporting iSCSI access for non-virtualised workloads, and using Storage Policy-Based management to name a few.


http://www.yellow-bricks.com/stickersshirts/Advanced, the next edition up the scale, gives benefits to vSAN platforms built on All-Flash. A vSAN host can be configured as “Hybrid” or “All-Flash”, hybrid uses spinning disks- HDD – for the capacity tier and flash disks for the cache tier. If your hardware uses the Hybrid model then there’s no advantage to using the Advanced edition as the additional features are only applicable to All-Flash configurations.

However, the Advanced edition provides major features which may tip the balance in the favour of the all-flash design when choosing a hardware platform, namely Erasure Coding, Deduplication, and Compression.

Deduplication (storing matching blocks of data only once), compression (using an algorithm to compress data), and erasure coding (basically arranging the data in a similar way to a RAID5/6 model but across hosts) are all features which can be used to reduce the amount of physical disk consumed by the data. Less space consumed = lower disk requirements = lower costs.  More detailed information on vSAN Space Efficiency can be found in the VMware docs.

As these three technologies can squash more data into your disk, they can provide more capacity for your spend and potentially offset the higher cost of SSD per gigabyte over HDD and the higher vSAN licensing price (Advanced is retailing at $1500 per socket more than Standard at time of writing). In the right circumstances it’s possible to design a higher capacity, better performing platform using the Advanced Edition+All-Flash route for the same cost of the Standard Edition+Hybrid design.

As with any data reduction technology, results can vary depending on workload. The VMware product page quotes savings of “up to” 7x from the deduplication and compression- in reality I’m seeing 2.5x in my own mixed server environment, but I’d expect VDI deployments to see higher figures. In my opinion All-Flash is the way forward and once you’ve made that decision then vSAN Advanced Edition wins over Standard in pretty much every situation.


The Enterprise edition takes the price up higher another $1500 per socket  (list price) at time of writing but adds Stretched Cluster with Local Failure Protection and Data-at-Rest Encryption.

The Encryption feature applies to whole vSAN cluster so everything stored on that datastore will be encrypted at rest. This allows dedupe and compression still to work, features that don’t provide any capacity benefits when using the alternative vSphere encryption. vSAN Encryption is configured using the same KMS service as vSphere- so a third-party Key Management Service is required, and you will want a non-encrypted (or alternatively encrypted) datastore/ cluster to host that on. As the encryption is a software offering expensive self-encrypting drives are not required.

The Stretched Cluster functionality allows a vSAN cluster to be designed to span across multiple datacentres and tolerate failure of an entire site. By using a witness host the cluster can detect the loss of a datacentre (or connectivity to it) and ensure that the storage is available using the alternative site. VMware High Availability ensures that any VMs which were running in the failed datacentre are powered on on the surviving site.



There are three editions of regular vSAN- Standard, Advanced, and Enterprise. If you want to use a Stretched vSAN Cluster or vSAN Encryption then you need the Enterprise edition. If you don’t expect to use either of these (perhaps vSphere Encryption covers any encryption requirements) and have an all-flash hardware configuration then it makes sense to go with Advanced Edition. And finally, if you have a Hybrid hardware config then go with Standard Edition (or upgrade your spinning disks to flash).

Further Reading:

vSAN Introduction

What is vSAN?

This is a first post in a series on vSAN, VMware’s software-defined-storage offering. vSAN uses disks within the ESXi hosts to create a resilient scalable shared storage platform for the virtual infrastructure providing many of the features of SAN/NAS shared storage without the need for additional hardware. There are also operational, performance, and scaling benefits associated with integrating the storage into the hypervisor’s control.

Unlike some other software-defined-storage solutions, vSAN is not a separate appliance but instead is baked into the ESXi hypervisor. In fact there is no separate install; installation (which I’ll cover in a future post) is simply a case of applying a license and configuring the cluster through vCenter.

vSAN dates back to 2013, with a General-Availability launch in March 2014 (the first was version 5.5 which was part of ESXi 5.5U1) and has evolved to the current version- v6.6 at the point this post was written.

Continue reading

VMworld Europe 2016 Day 3

Wednesday morning and VMworld continues…

Here’s a look at my sessions from another day in Barcelona.

General Session


The big announcements of Day 2 gave way to some more detail in the Day 3 Keynote, kicked off by Sanjay Poonen focusing on the digital transformation happening in the EUC world. VMware has 66,000 EUC customers and is promoting an any-app, any-device strategy through the Workspace One platform. Today 50% of business apps are web based, 40% are Windows Client-Server, and Mobile picks up the remainder. We saw a demo of how VMware use workspace one internally to provide access to all these services through one portal.

Sanjay was followed on stage by Ray O’Farrell who gave us further detail about the announcements in the SDDC. The vSphere 6.5 announcement was of course at the forefront, with Ray pointing out the advances that made the interface six times faster than 5.5 and gave a demo using PowerCLI which showed the realtime response of the new HTML5 client. He ran through a demo of the new VM encryption facilities in 6.5 and covered some of the advances in vRealize automation- including new support for containers.

imageNext up was Yanbing Li to talk about vSAN (lower case v, upper case SAN) which also has some advantages in the 6.5 release. First up is Direct-Connect options for 2-node deployments (think remote office/ branch office installs or setups requiring strict isolation of data). In this model the management and witness traffic is split out from the data traffic allowing for the two vSAN nodes to be linked directly together using Ethernet cables.

Secondly, vSAN 6.5 enables iSCSI support, so the storage infrastructure can now be used as a target for physical workloads. Yanbing also talked about the ongoing vSAN beta- future versions of the product are likely to offer data at-rest encryption and nested fault domains. vSAN has passed 5000 customers in 2.5 years, and now VMware hopes more affordable licensing with the offer of all-flash support on that Standard SKU.

Networking and Security was the next topic with Rajiv Ramaswami with the focus on NSX which is now giving 1700 NSX customers Micro-segmentation and Distributed Load Balancing. From my point of view NSX is continuing to gain weight as the product underpinning VMware’s SDDC- most of the presentations this week mentioned NSX in some form – and I expect this market to grow as the Private/Public/Hybrid cloud model expands.

Kit Colbert brought the Wednesday Keynote to a close with insights into container hosting and management VMware style using the Photon Platform. As with everything else here Photon is continuing to evolve and will be offering Kubernetes-as-a-service in Q4 this year. The Photon Controller and Photon OS are both open source- available for download from Github.

If you want to watch the full session yourself, check out the video here:

Day 2 Operations: A vCenter Server Administrator’s Diary [INF9128]

Adam Eckerle and Emad Younis gave this talk, catching up on what’s new in vCenter and how to keep it running smoothly after the install process has finished. I picked up some great takeaways here, and I’ve distilled my pages of notes to come up with the following highlights:

  • There are 5 web based clients for the vSphere environment in 6.5 :  vCenter Client and Web Client, the Appliance Manager UI (formerly VAMI) and so on. There is no support for the legacy Windows C# client.
  • The vCenter appliance upgrade preserves the identity of the old Windows-based vCenter so all connected applications and plugins should continue to operate. If the upgrade needs to be rolled back it’s just a case of turning off and removing the new VCSA and then powering on the old Windows Server and rejoining it to the domain (although any configuration changes made under the VCSA’s stewardship would then be lost) . Upgrades are possible from Windows vCenter 5.5 or 6.0
  • We were shown a demo of extending the disk in the vCenter appliance using LVM autogrow. Also, as of 6.5, the appliance will warn when the disk reaches 80% capacity and will auto-shutdown at 95% to prevent corruption

LVM Autogrow for logs on VCSA

  • The second half of the session included a whiteboarded overview of PSC migrations and topology. The linear ring topology was highlighted as being preferable over hub/spoke (limited in failover) and full mesh(too complicated) in larger (and expandable) deployments. Check out the poster for more details on choosing a PSC topology.

vBrownBag “The Amazing World of IT in Higher Education”

This time it was my turn to give a presentation, and in this quick ten minute talk at the vBrownBag stage I covered some of the unusual practices that an IT Pro might experience if dropped into a University environment and how various forms virtualisation can be used to save the day. If you want some more details check out my post and video link here.

VMUG Banner

London VMUG April 2016 (Part 1)

Thursday saw another trip to TechUK for the second London VMware Users Group of the year. It was, yet again, a very educational and entertaining community driven event with sessions covering a whole range of topics from enterprise scale implementations of hardware and software right down to the nitty-gritty of running your own HomeLab.

After the initial welcomes and introductions, the day kicked of with a session from Luca Dell’Oca of Veeam. Entitled “Veeam Backup and Replication: Worst Practices”, this was an often humorous look at what not to do when planning and implementing a backup regime. He covered many of the common mistakes when backing up a virtual environment, and although some parts were product specific (for example how the Veeam scheduler prior to v9 was not optimised for the hundreds of jobs a per-VM backup strategy would involve) there was a lot of generic advice. Highlights for me included

  • Remember that in-guest iSCSI connections will not be touched by VMware snapshots so don’t forget about them when backing up the VMs.
  • Use monitoring tools- in particular keep a close eye on storage space used by snapshots and performance. vCenter performance is critical and can often be a bottleneck.
  • When are other infrastructure tasks scheduled? Don’t run backups when your vCenter SQL maintenance plan is running for example.
  • Find the balance between one backup job per VM and one backup job for all VMs. Group VMs together based on backup policy.
  • Don’t be a cheapskate! Fast, Good, and Cheap is not possible.
  • It’s not necessary to fiddle with every advanced setting. Don’t change what you don’t understand!

Next on the agenda was the Plenary Keynote, where Simon Richardson of VMware went through the new developments in the Software Defined Datacentre portfolio. He pointed out that the only constant is change, and that means there’s always a lot to catch up on.

Simon discussed using the Hyperconverged SDDC as the best of both the traditional DC world and the Google/Facebook/Amazon “extreme” style of datacentre. Applying the analogy of a water company to IT he suggested we want to be in the position where we can turn on a tap and compute comes out.

There’s new features in the vRealize suite as well, vROps producing intelligent workload balancing and Log Insight for diving into the cause of a fault or performance issue.

vRealize Business was a new component for me (more on that later in the day) and I can definitely see the use of being able to show (or chargeback) the actual cost of a VM- taking into account all the factors- environment, hardware, software, support, manpower, and so on. Although there was little OpenStack adoption in the audience, Simon also touched on VIO- VMware Integrated OpenStack.

After a quick break in the Thames Suite (named in honour the breakroom in the previous London VMUG venue 🙂 ) I caught up with VSAN in more depth in a talk by Simon Todd (and earned some Cloud Credibility points along the way). VMware’s Virtual SAN is getting wider and wider adoption, even branching outside of the traditional Datacentre environment into Offshore Oil Rigs and even Submarines. This is more than just Business critical this really is Mission critical. The ability to deploy (and scale) large arrays quickly is a key selling point, multipetabyte arrays spun up in a matter of minutes and customers who had cut both costs and deployment times.

@MrVirtualSAN at @LonVMUG

What was my vmdk on VSAN up to at 4am last night? A user complained their service was slow.

Moving a bit more into the technical detail- Simon explained how VSAN is an object store with each VMDK, each file, becoming an object. This allows for per-object policy, so performance, IOPS limits, and resilience can be set, and then monitored, at the VMDK level. When pushing for that ultimate performance, this granularity could be of great benefit.

That brought us to the halfway point in the day’s proceedings, where there was lunch an and opportunity to catch up with the events sponsors- PernixData, Veeam, and Nimble Storage.

London VMUG dates

Upcoming London VMUG dates

Edit- Part 2 can be found here.