Tag Archives: contention

Example Architectural Decision – Transparent Page Sharing (TPS) Configuration for VDI (2 of 2)

Posted on by
Reply

Problem Statement

In a VMware vSphere environment, with future releases of ESXi disabling Transparent Page Sharing by default, what is the most suitable TPS configuration for a Virtual Desktop environment?

Assumptions

1. TPS is disabled by default
2. Storage is expensive
3. Two Socket ESXi Hosts have been chosen to align with a scale out methodology.
4. Average VDI user is Task Worker with 1vCPU and 2GB Ram.
5. Memory is the first compute level constraint.
6. HA Admission Control policy used is “Percentage of Cluster Resources reserved for HA”
7. vSphere 5.5 or earlier

Requirements

1. VDI environment costs must be minimized

Motivation

1. Reduce complexity where possible.
2. Maximize the efficiency of the infrastructure

Architectural Decision

Enable TPS and disable Large Memory pages

Justification

1. Disabling Large pages is essential to maximizing the benefits of TPS
2. Not disabling large pages would likely result in minimal TPS savings
3. With Kiosk and Task worker VDI profiles, the percentage of memory which is likely to be shared is higher than for Power users.
4. Existing shared storage has plenty of spare Tier 1 capacity to vSwap files

Implications

1. Sufficient capacity for VM swap files must be catered for.
2. VDI & Storage performance may be impacted significantly in the event of memory contention.
3. Decreased memory costs may result in increased storage costs.
4. During patching, and operational verification that non default settings have not been reverted by the patching of ESXi.
5. Additional CPU overhead on ESXi from enabling TPS.
6. HA admission control will calculate fail-over requirements (when using Percentage of cluster resources reserved for HA) so that performance will be approximately the same in the event of a fail-over due to reserving the full RAM reserved for every VM,
6. HA admission control (when configured to Percentage of Cluster resources reserved for HA) will only calculate fail-over capacity based on 0MB + VM overhead for each VM which can lead to significantly degraded performance in a HA event.
7. Higher core count (and higher cost) CPUs may be desired to drive overcommitment ratios as RAM will be less likely to be a point of contention.

Alternatives

1. Use 100% memory reservation and leave TPS disabled (default)
2. Use 50% memory reservation and Enable TPS and disable large pages

Related Articles:

1. The Impact of Transparent Page Sharing (TPS) being disabled by default @josh_odgers (VCDX#90)

2. Example Architectural Decision – Transparent Page Sharing (TPS) Configuration for VDI (1 of 2)

3. Future direction of disabling TPS by default and its impact on capacity planning –@FrankDenneman (VCDX #29)

4. Transparent Page Sharing Vulnerable, Yet Largely Irrelevant – @ChrisWahl(VCDX#104)

The future of NAS Storage (NFS) for Virtual Environments

Posted on by
9

I read the article (below) by Howard Marks after seeing it come up on Twitter today, and I found it to be very interesting and refreshing to read as it hits the nail on the head.

http://www.networkcomputing.com/storage-networking-management/vmware-has-to-step-up-on-nfs/240163350

For a long time Network Attached Storage (NAS) has been considered by many (including myself in the past) as a second class citizen, or Tier 3 storage and not a serious choice for mission critical virtual environments.

In recent years, I have used more and more NFS in vSphere environments, and as I went through my VCDX journey I formed the strong view that NFS was in fact the best storage protocol for vSphere/vCloud/View environments having gone through a process of trying to learn as much as possible about every storage alternative available to vSphere.

In fact my VCDX design was based on a vCloud solution running on NFS, and this was one area I found quite easy to present and defend during the VCDX defence due to the many advantages of NFS.

In the article, Howard wrote

“It’s time for VMware to upgrade its support for file storage (as opposed to block storage) and embrace the pioneering vendors who are building storage systems specifically for the virtualization environment.”

I totally agree with this statement, and I think it is in the best interest of VMware, its partners and customers for VMware to go down this path. I think most would agree that Netapp have been leading the charge with NFS based storage for a long time, and in my opinion rightly so, with some new storage vendors also choosing to build solutions around NFS.

Another comment Howard made was

“managing vSphere with NFS storage is somewhat simpler than managing an equivalent system on block storage. Even better, a good NFS storage system, because it knows which blocks belong to which virtual machine, can perform storage management tasks such as snapshots, replication and storage quality of service per virtual machine, rather than per volume.”

I totally agree with the above statement and VMware’s development of features such as View Composer for Array Integration (VCAI) which is only supported on NFS, shows the protocol has significant advantages over block based storage especially for deployment speed and reduced workload on the storage compared. (VCAI uses the Fast File Clone VAAI-NAS Primitive to create near instant space efficient Linked Clone desktops)

I wrote an example architectural decision regarding storage protocol choice for Horizon View (VDI) environments which covers in more depth the advantages of NFS for VDI environments. The article can be viewed here : Example Architectural Decision – Storage Protocol Choice for Horizon View

Also NFS does not suffer from the same challenges as block based storage, as much larger numbers of VMs can share an NFS datastore compared to VMFS datastore without being negatively impacted by latency as a result of SCSI reservations (although vastly improved with VAAI) or contention resulting from limited SCSI queue depths which is something VAAI does still not address.

These limitations of block storage leads to the number of VMs per datastore remaining at the old rules of thumb of <25 for non I/O intensive workloads even with VAAI which some felt was the magic solution to the issue which sadly was incorrect. (Note: Number of desktop VMs per VMFS datastore with VAAI the recommended maximum is 140 compared to 64 without VAAI and NFS of >200).

Howard went on to write

“The first step would be for VMware to acknowledge that NFS has advanced in the past decade.”

I think this has been acknowledged by VMware along with many experts in the industry which is a positive step forward and I believe VMware will give more attention to NFS in future versions.

Howard further commented that

“Today vSphere supports version 3.0 of NFS—which is seventeen years old. NFS 4.1 has much more sophisticated security, locking and network improvements than NFS 3.0. The optional pNFS extension can bring the performance and multipathing of SANs with centralized file system management.”

I really think that VMware adding support in the future for NFS 4.1 will really help cement NFS as the protocol of choice for virtual environments and will be complimentary to VMware’s upcoming VSAN offering.

I think with bolstered NFS support and VSAN, VMware will have a solid storage layer to take virtualization into the future, without requiring storage vendors to immediately support vVOLs which in my opinion is being built (at least in part) to solve the challenges of VMFS and block based storage, when NFS (even version 3.) addresses most requirements in virtual environments very well today, and NFS 4.1 support will only improve the situation.

Howard’s comment (below) appears to echo these thoughts.

“Better NFS support will empower storage vendors to innovate and strengthen the vSphere ecosystem and fill the gap until vVols are ready. NFS support will also provide an alternative once vVols hit the market.”

 

To finish I thought Howard’s comment on snapshots (below) and replication being per Virtual Machine rather than volume or LUN, several vendors are doing this today moving towards NFS 4.1 will help these vendors continue to innovate and provide better and more efficient storage solutions for VMware’s customers which I think is what everyone wants.

Even better, a good NFS storage system, because it knows which blocks belong to which virtual machine, can perform storage management tasks such as snapshots, replication and storage quality of service per virtual machine, rather than per volume.

Example Architectural Decision – Datastore (LUN) Sizing with Block Based Storage

Posted on by
2

Problem Statement

In a vSphere environment, What is the most suitable Datastore (LUN) sizing to use for to support both production & development workloads to ensure minimum storage overhead and optimal performance?

Requirements

1. RTO 4hrs
2. RPO 12hrs
3. Support Production and Test & Development Workloads
4. Ensure optimal storage capacity utilization
5. Ensure storage performance is both consistent & maximized
6. Ensure the solution is fully supported
7. Minimize BAU effort (Monitoring)

Assumptions

1. Business critical applications are excluded
2. Block based storage
3. VAAI is supported and enabled
4. VADP backups are being utilized
5. vSphere 5.0 or later
6. Storage DRS will not be used
7. SRM is in use
8. LUNs & VMs will be thin provisioned
9. Average size VM will be 100GB and be 50% utilized
10. Virtual machine snapshot will be used but not for > 24 hours
11. Change rate of average VM is <= 15% per 24 hour period
12. Average VM has 4GB Ram
13. No Memory reservations are being used
14. Storage I/O Control (SOIC) is not being used
15. Under normal circumstances storage will not be over committed at the storage array level.
16. The average maximum IOPS per VMs is 125 (16Kb) (MBps per VM <=2)
17. The underlying storage has sufficient performance to cater for the average maximum IOPS per VM
18. A separate swap file datastore will be configured per cluster

Constraints

1. Must used existing storage solution (Block Based Storage)

Motivation

1. Increase flexibility
2. Ensure physical disk space is not unnecessarily wasted
3. Create a Scalable solution
4. Ensure high performance
5. Ensure high utilization of storage resources by reducing “islands” of unused capacity
6. Provide flexibility in the unit size of partial SRM failovers

Architectural Decision

The standard datastore size will be 3TB and contain up to 25 standard virtual machines.

This is based on the following

25 VMs per datastore X 100GB (Assumes no over commitment) = 2500GB

25 VMs w/ 4GB RAM = 100GB minus 0Gb reservation = 100GB vswap space to be stored on the swap file datastore

25 VMs w/ Snapshots of up to 15% =  375GB

Total = 2500GB + 375GB = 2875GB

Average capacity used per VM = 115GB

Justification

1. In worst case scenario where every VM has used 100% of its VMDK capacity and has 4GB RAM with no memory reservation and a snapshot of up to 15% of its size the 3TB datastore will still have 197GB remaining, as such it will not run out of space.
2. The Queue depth is on a per datastore (LUN) basis, as such, having 25 VMs per LUNs allows for a minimum of 1.28 concurrent I/O operations per VM based on the standard queue depth of 32 although it is unlikely all VMs will have concurrent I/O so the average will be much higher.
3. Thin Provisioning minimizes the impact of situations where customers demand a lot of disk space up front when they only end up using a small portion of the available disk space
4. Using Thin provisioning for VMs increases flexibility as all unused capacity of virtual machines remains available on the Datastore (LUN).
5. VAAI automatically raises an alarm in vSphere if a Thin Provisioned datastore usage is at >= 75% of its capacity
6. The impact of SCSI reservations causing performance issues (increased latency) when thin provisioned virtual machines (VMDKs) grow is unlikely to be an issue for 25 low I/O VMs and with VAAI is no longer an issue as the Atomic Test & Set (ATS) primitive alleviates the issue of SCSI reservations.
7. As the VMs are low I/O it is unlikely that there will be any significant contention for the queue depth with only 25 VMs per datastore
8. The VAAI UNMAP primitive provides automated space reclamation to reduce wasted space from files or VMs being deleted
9. Virtual machines will be Thin provisioned for flexibility, however they can also be made Thick provisioned as the sizing of the datastore (LUN) caters for worst case scenario of 100% utilization while maintaining free space.
10. Having <=25 VMs per datastore (LUN) allows for more granular SRM fail-over (datastore groups)

Alternatives

1.  Use larger Datastores (LUNs) with more VMs per datastore
2.  Use smaller Datastores (LUNs) with less VMs per datastore

Implications

1. When performing a SRM fail over, the most granular fail over unit is a single datastore which may contain up to 25 Virtual machines.

2. The solution (day 1) does not provide CapEx saving on disk capacity but will allow (if desired) over commitment in the future

Thanks to James Wirth (VCDX#83) @JimmyWally81 for his contributions to this example decision.

Related Articles

1. Datastore (LUN) and Virtual Disk Provisioning (Thin on Thick)

2. Datastore (LUN) and Virtual Disk Provisioning (Thin on Thin)

3. Virtual Machine vSwap Location

CloudXClogo