Shulou(Shulou.com)05/31 Report--

This article introduces you how to Virtual SAN/VSAN FAQ, the content is very detailed, interested friends can refer to, hope to be helpful to you.

Q: VSAN needs to use SSD. What is its use?

Answer: SSD is used for read cache (70%) and write buffer (30%). Each write will first go to SSD, and then cancel the temporary save to HDD later.

Q: when creating a VSAN virtual machine storage policy, when should I use "allowed number of failures" and when should I use "stripe width"?

A: the number of failures allowed is related to availability, and you can define this policy if you need to keep virtual machines available in the event of a host or disk group failure. Therefore, if 1 host failure is allowed, the policy can be defined as 1. In this way, 2 data objects and 1 witness are created in the cluster. "stripe width" is related to performance (that is, read performance when not in cache and write staging canceled). A value of 2 or higher stripes the data across multiple disks. If used in conjunction with "allowed number of failures", data in a virtual machine may be stored on multiple disks on multiple hosts.

Q: does VSAN have a default storage policy?

A: yes, a default policy is applied to all virtual machines on the VSAN data store, but you don't see it in vSphere UI. Using the following command, you can see that the default policy is defined as a different class: esxcli vsan policy getdefault. The "allow 1 failure number" policy is applied by default, so that the object is resilient even if the user forgets to create and set the policy. It is not recommended to change the default policy.

Q: how is the data striped across multiple disks on the host when the stripe width is set to 2?

A: first of all, setting the stripe width to 2 does not guarantee that the data will be striped on multiple disks on a host. VSAN's own algorithm determines where data should be placed, so although all hosts have enough disks, data may still be striped across multiple hosts rather than multiple disks on a host. Data striping is done in blocks of 1MB.

Q: since VSAN creates a data store, what is the purpose of "disk groups"?

A: disk groups define SSD,SSD in front of a set of HDD for caching / buffering. Generally speaking, disk groups are a way to map HDD to SSD. Each disk group has 1 SSD and up to 6 disks. Based on customer feedback, we plan to increase the number of HDD to 7 so that a disk group can be built on a host with 8 physical disk slots.

Q: how many disks can a host provide for VSAN?

Answer: a maximum of 5 disk groups; each disk group requires 1 SDD and a minimum of 1 and a maximum of 6 HDD. The maximum number of HDD per host is 5 x 6 = 30. The maximum number of SSD per host is 5 x 1 = 5.

Q: can I support both SSD and PCIe flash cards?

A: yes, you can support both. However, check the HCL for more information, as there are some guidelines and requirements to pay attention to.

Q: does VSAN have to use 10GbE?

A: VSAN does not require the use of 10GbE. VSAN can also work well with 1GbE in smaller environments, such as in the lab. Note that 10GbE is a recommended value.

Q: why is it recommended that the isolated response for HA be configured as "powered-off"?

A: when VSAN is enabled, vSphere HA uses the VSAN VMkernel network to detect signals. If the host does not receive any detection signals, it is likely that the host has been isolated / partitioned from the rest of the cluster from the perspective of VSAN. In this case, it is recommended that you power off the virtual machine because HA automatically powers on a new copy on the remaining hosts in the cluster. In this way, after the host is out of isolation, there will not be two virtual machines with the same identity in the network.

Q: can I partition SSD or disk and use them for other purposes (for example, installing ESXi/vFlash)?

Answer: no, you cannot partition SSD or HDD. Virtual SAN always takes up the entire disk. Perhaps the most feasible way to use VSAN is to install ESXi on the internal USB/SD card, which maximizes the capacity of VSAN.

Q: does VSAN support repetition or compression?

Answer: the current version of VSAN does not support duplication or compression. In the VSAN cluster, the most expensive resource is SSD/ flash, so data duplication is most likely to occur at this tier. "if there are multiple copies of data, there will be two replicas on the HDD, two temporary replicas on the distributed write buffer (30 percent of SSD), and the distributed read cache portion of flash memory (70 percent) contains only one copy for any cached data."

Q: can VSAN take advantage of SAN/NAS data storage?

A: VSAN currently does not support the use of SAN/NAS data storage. The disk must be "local" and passed directly to the host.

Q: some people say that if I set Object Space Reservation (object space reservation) to 100%, VSAN will simplify the disk by default. Does this mean that VMDK will become a thick zero setting?

A: no. The definition of "Object Space Reservation" does not mean that the virtual machine is provided with thick or partial thick provisioning. If you calculate using used disk space / free disk space, the object space reservation is entirely related to the amount used by VSAN. If you set Object Space Reservation (object space reservation) to 100% on a 25GB disk, the disk will become a thin standby disk, but VSAN will assume that the 25GB has 100% used space to calculate. I think it can be compared with memory reservation.

Q: will VSAN use iSCSI or NFS to connect hosts to the data store?

A: VSAN does not use any of these to connect hosts to the data store. It uses a proprietary mechanism.

Q: what is the impact of maintenance mode in VSAN-enabled clusters?

A: hosts that provide storage for VSAN data stores can be placed in maintenance mode in three ways:

1) full data migration-migrate all data on the host. Impact: it may take a long time to complete.

2) ensure accessibility-VSAN keeps all virtual machines accessible at all times by migrating the necessary data to other hosts. Impact: usability policies may be violated.

3) No data migration-No data is migrated. Impact: depending on the defined "number of failures allowed" policy, some virtual machines may be unavailable.

The safest option is option 1, and option 2 is the preferred and default option because it is the fastest to complete. I think the question is why the host is put in maintenance mode and how long the host can be used again. If you do need to enter maintenance mode quickly and do not care about the possibility of data loss, you can select option 3: fallback.

Q: does vSphere have features that VSAN does not support or are not compatible with?

A: currently, VSAN does not support vSphere Distributed Power Management, Storage DRS, and Storage IO Control.

Q: how do I add a Virtual SAN/VSAN license?

A: the VSAN license is applied at the cluster level. Open Web Client, click a cluster that has VSAN enabled, and then click the Manage tab and then click Settings. Under Configuration, click Virtual SAN Licensing (Virtual SAN license), and then click Assign License Key (assign license key).

Q: how is Virtual SAN priced / licensed?

A: VSAN will license on a per-slot basis, but pricing details have not been made public. Note that the current VSAN Beta license code includes DVS and virtual machine storage policies, even with vSphere licenses lower than Enterprise Plus.

Q: if the host fails, resulting in data loss, and all virtual machines are protected by the Number1 policy, how long will it take for VSAN to start rebuilding the lost data?

A: VSAN determines which objects (that is, objects protected by the Number1 policy and stored on the host) are not compliant and initiates a 60-minute timeout period. This timeout period is to avoid unnecessary and costly full data synchronization. If the host recovers within this 60 minutes, the system replicates the resulting differences to the host. "if the virtual machine has multiple mirrors and does not notice the failure, after this 60-minute period, it returns to full policy compliance to cope with the impact of more failures."

Q: when a virtual machine moves in a cluster, will its objects move with it to keep the IO local?

A: no, objects, such as virtual disks, will not move with the virtual machine. Just imagine the cost / overhead of moving virtual disks between hosts each time DRS recommends a migration. At this point, IO can be executed remotely. This means that while a virtual machine may be running on host 1 from an CPU/ memory perspective, its virtual disks may actually be on host 2 and host 3.

Q: after a virtual machine is migrated to another host, will the SDD cache (which temporarily affects performance) be lost if vMotion is performed, and will the cache be gradually rebuilt?

A: no cache is lost, and there is no need to rebuild / re-hot enable the cache. If necessary, the cache can be accessed remotely.

Q: does VSAN support Fault Tolerance (i.e. FT)?

Answer: no, this version of VSAN does not support Fault Tolerance.

Q: in Sphere, the SSD in the host is reported as "non-SSD". According to technical support, this issue is a known problem with the version of the server being used. Will disk type "false positives" affect the configuration of VSAN?

A: yes. You need to use the relevant identifiers to mark the SSD as local (the following example is what I used in the lab and may be different from yours). Here, I set it to "local" and "SSD".

Esxcli storage nmp satp rule add-satp VMW_SATP_LOCAL-device mpx.vmhba2:C0:T0:L0-option "enable_local enable_ssd"

Q: as mentioned above, after a failure, it will take 60 minutes for VSAN to start repairing automatically. Can you shorten this timeout value?

A: disclaimer: it is not recommended to change this value, and I am not sure if I support such a change

Yes, you can configure an advanced setting named "VSAN.ClomRepairDelay" on each host in the VSAN cluster to shorten the timeout value.

Q: why can't the data store detection signal feature be used in a cluster with only VSAN?

A: there is no requirement for detection signal data storage. This feature cannot be used when only VSAN data is stored because HA uses the VSAN network to detect signals. Therefore, if the host is isolated from the VSAN network and cannot send a detection signal, it is safe to say that the host cannot update the detection signal area remotely, so it makes no sense to enable this feature in an environment with only VSAN.

Q: are there any specific best practices for deploying View on VSAN?

A: yes, it is mainly for availability / caching and capacity reservation. Andre Leibovici has written a good article on this topic, so read it now! (http://myvirtualcloud.net/?p=5440)

Q: can the VSAN VMkernel of hosts in one cluster belong to another subnet?

A: VSAN VMkernel must belong to the same subnet. It is not supported that one (or more) hosts in a VSAN cluster are on different subnets. If you use multiple VMkernel interfaces for each host, each interface must belong to a different subnet!

Q: does VSAN support spanning multiple geographic locations?

Answer: the current version of VSAN does not support Metro clusters.

Q: what's the difference between a host failure and a gradual disk failure?

A: there is indeed a difference. There are a variety of failure states, based on which you can determine how quickly VSAN starts a new mirror. Two of the failure states are "do not exist" and "degraded". Degraded refers to a disk failure that has been recognized by the system and knows that the disk has not been restored. In this case, VSAN recognizes the "degraded" state and immediately creates a new mirror for the affected object, because since it knows that the disk will not recover soon, it does not have to wait for 60 minutes. The "non-existent" state means that VSAN does not know whether the disk will recover soon, either because the host fails or suddenly moves the disk, and the system starts with a 60-minute timeout.

Q: what happens if SSD fails in a VSAN cluster?

A: SSD is in front of the disk group and acts as a read cache / write buffer. If the SSD fails, the disk group and all its stored components are marked as degraded. Then, if there is enough disk capacity, VSAN creates a new mirrored copy where appropriate. For more information, read http://www.yellow-bricks.com/2013/09/18/vsan-handles-disk-host-failure/

Q: does vSphere support TRIM for SSD?

A: no, TRIM is not currently supported, and TRIM will not be used.

Q: VSAN needs to use SSD. What is its use?

Answer: SSD is used for read cache (70%) and write buffer (30%). Each write will first go to SSD, and then cancel the temporary save to HDD later.

Q: when creating a VSAN virtual machine storage policy, when should I use "allowed number of failures" and when should I use "stripe width"?

A: the number of failures allowed is related to availability, and you can define this policy if you need to keep virtual machines available in the event of a host or disk group failure. Therefore, if 1 host failure is allowed, the policy can be defined as 1. In this way, 2 data objects and 1 witness are created in the cluster. "stripe width" is related to performance (that is, read performance when not in cache and write staging canceled). A value of 2 or higher stripes the data across multiple disks. If used in conjunction with "allowed number of failures", data in a virtual machine may be stored on multiple disks on multiple hosts.

Q: does VSAN have a default storage policy?

A: yes, a default policy is applied to all virtual machines on the VSAN data store, but you don't see it in vSphere UI. Using the following command, you can see that the default policy is defined as a different class: esxcli vsan policy getdefault. The "allow 1 failure number" policy is applied by default, so that the object is resilient even if the user forgets to create and set the policy. It is not recommended to change the default policy.

Q: how is the data striped across multiple disks on the host when the stripe width is set to 2?

A: first of all, setting the stripe width to 2 does not guarantee that the data will be striped on multiple disks on a host. VSAN's own algorithm determines where data should be placed, so although all hosts have enough disks, data may still be striped across multiple hosts rather than multiple disks on a host. Data striping is done in blocks of 1MB.

Q: since VSAN creates a data store, what is the purpose of "disk groups"?

A: disk groups define SSD,SSD in front of a set of HDD for caching / buffering. Generally speaking, disk groups are a way to map HDD to SSD. Each disk group has 1 SSD and up to 6 disks. Based on customer feedback, we plan to increase the number of HDD to 7 so that a disk group can be built on a host with 8 physical disk slots.

Q: how many disks can a host provide for VSAN?

Answer: a maximum of 5 disk groups; each disk group requires 1 SDD and a minimum of 1 and a maximum of 6 HDD. The maximum number of HDD per host is 5 x 6 = 30. The maximum number of SSD per host is 5 x 1 = 5.

Q: can I support both SSD and PCIe flash cards?

A: yes, you can support both. However, check the HCL for more information, as there are some guidelines and requirements to pay attention to.

Q: does VSAN have to use 10GbE?

A: VSAN does not require the use of 10GbE. VSAN can also work well with 1GbE in smaller environments, such as in the lab. Note that 10GbE is a recommended value.

Q: why is it recommended that the isolated response for HA be configured as "powered-off"?

A: when VSAN is enabled, vSphere HA uses the VSAN VMkernel network to detect signals. If the host does not receive any detection signals, it is likely that the host has been isolated / partitioned from the rest of the cluster from the perspective of VSAN. In this case, it is recommended that you power off the virtual machine because HA automatically powers on a new copy on the remaining hosts in the cluster. In this way, after the host is out of isolation, there will not be two virtual machines with the same identity in the network.

Q: can I partition SSD or disk and use them for other purposes (for example, installing ESXi/vFlash)?

Answer: no, you cannot partition SSD or HDD. Virtual SAN always takes up the entire disk. Perhaps the most feasible way to use VSAN is to install ESXi on the internal USB/SD card, which maximizes the capacity of VSAN.

Q: does VSAN support repetition or compression?

Answer: the current version of VSAN does not support duplication or compression. In the VSAN cluster, the most expensive resource is SSD/ flash, so data duplication is most likely to occur at this tier. "if there are multiple copies of data, there will be two replicas on the HDD, two temporary replicas on the distributed write buffer (30 percent of SSD), and the distributed read cache portion of flash memory (70 percent) contains only one copy for any cached data."

Q: can VSAN take advantage of SAN/NAS data storage?

A: VSAN currently does not support the use of SAN/NAS data storage. The disk must be "local" and passed directly to the host.

Q: some people say that if I set Object Space Reservation (object space reservation) to 100%, VSAN will simplify the disk by default. Does this mean that VMDK will become a thick zero setting?

A: no. The definition of "Object Space Reservation" does not mean that the virtual machine is provided with thick or partial thick provisioning. If you calculate using used disk space / free disk space, the object space reservation is entirely related to the amount used by VSAN. If you set Object Space Reservation (object space reservation) to 100% on a 25GB disk, the disk will become a thin standby disk, but VSAN will assume that the 25GB has 100% used space to calculate. I think it can be compared with memory reservation.

Q: will VSAN use iSCSI or NFS to connect hosts to the data store?

A: VSAN does not use any of these to connect hosts to the data store. It uses a proprietary mechanism.

Q: what is the impact of maintenance mode in VSAN-enabled clusters?

A: hosts that provide storage for VSAN data stores can be placed in maintenance mode in three ways:

1) full data migration-migrate all data on the host. Impact: it may take a long time to complete.

2) ensure accessibility-VSAN keeps all virtual machines accessible at all times by migrating the necessary data to other hosts. Impact: usability policies may be violated.

3) No data migration-No data is migrated. Impact: depending on the defined "number of failures allowed" policy, some virtual machines may be unavailable.

The safest option is option 1, and option 2 is the preferred and default option because it is the fastest to complete. I think the question is why the host is put in maintenance mode and how long the host can be used again. If you do need to enter maintenance mode quickly and do not care about the possibility of data loss, you can select option 3: fallback.

Q: does vSphere have features that VSAN does not support or are not compatible with?

A: currently, VSAN does not support vSphere Distributed Power Management, Storage DRS, and Storage IO Control.

Q: how do I add a Virtual SAN/VSAN license?

A: the VSAN license is applied at the cluster level. Open Web Client, click a cluster that has VSAN enabled, and then click the Manage tab and then click Settings. Under Configuration, click Virtual SAN Licensing (Virtual SAN license), and then click Assign License Key (assign license key).

Q: how is Virtual SAN priced / licensed?

A: VSAN will license on a per-slot basis, but pricing details have not been made public. Note that the current VSAN Beta license code includes DVS and virtual machine storage policies, even with vSphere licenses lower than Enterprise Plus.

Q: if the host fails, resulting in data loss, and all virtual machines are protected by the Number1 policy, how long will it take for VSAN to start rebuilding the lost data?

A: VSAN determines which objects (that is, objects protected by the Number1 policy and stored on the host) are not compliant and initiates a 60-minute timeout period. This timeout period is to avoid unnecessary and costly full data synchronization. If the host recovers within this 60 minutes, the system replicates the resulting differences to the host. "if the virtual machine has multiple mirrors and does not notice the failure, after this 60-minute period, it returns to full policy compliance to cope with the impact of more failures."

Q: when a virtual machine moves in a cluster, will its objects move with it to keep the IO local?

A: no, objects, such as virtual disks, will not move with the virtual machine. Just imagine the cost / overhead of moving virtual disks between hosts each time DRS recommends a migration. At this point, IO can be executed remotely. This means that while a virtual machine may be running on host 1 from an CPU/ memory perspective, its virtual disks may actually be on host 2 and host 3.

Q: after a virtual machine is migrated to another host, will the SDD cache (which temporarily affects performance) be lost if vMotion is performed, and will the cache be gradually rebuilt?

A: no cache is lost, and there is no need to rebuild / re-hot enable the cache. If necessary, the cache can be accessed remotely.

Q: does VSAN support Fault Tolerance (i.e. FT)?

Answer: no, this version of VSAN does not support Fault Tolerance.

Q: in Sphere, the SSD in the host is reported as "non-SSD". According to technical support, this issue is a known problem with the version of the server being used. Will disk type "false positives" affect the configuration of VSAN?

A: yes. You need to use the relevant identifiers to mark the SSD as local (the following example is what I used in the lab and may be different from yours). Here, I set it to "local" and "SSD".

Esxcli storage nmp satp rule add-satp VMW_SATP_LOCAL-device mpx.vmhba2:C0:T0:L0-option "enable_local enable_ssd"

Q: as mentioned above, after a failure, it will take 60 minutes for VSAN to start repairing automatically. Can you shorten this timeout value?

A: disclaimer: it is not recommended to change this value, and I am not sure if I support such a change

Yes, you can configure an advanced setting named "VSAN.ClomRepairDelay" on each host in the VSAN cluster to shorten the timeout value.

Q: why can't the data store detection signal feature be used in a cluster with only VSAN?

A: there is no requirement for detection signal data storage. This feature cannot be used when only VSAN data is stored because HA uses the VSAN network to detect signals. Therefore, if the host is isolated from the VSAN network and cannot send a detection signal, it is safe to say that the host cannot update the detection signal area remotely, so it makes no sense to enable this feature in an environment with only VSAN.

Q: are there any specific best practices for deploying View on VSAN?

A: yes, it is mainly for availability / caching and capacity reservation. Andre Leibovici has written a good article on this topic, so read it now! (http://myvirtualcloud.net/?p=5440)

Q: can the VSAN VMkernel of hosts in one cluster belong to another subnet?

A: VSAN VMkernel must belong to the same subnet. It is not supported that one (or more) hosts in a VSAN cluster are on different subnets. If you use multiple VMkernel interfaces for each host, each interface must belong to a different subnet!

Q: does VSAN support spanning multiple geographic locations?

Answer: the current version of VSAN does not support Metro clusters.

Q: what's the difference between a host failure and a gradual disk failure?

A: there is indeed a difference. There are a variety of failure states, based on which you can determine how quickly VSAN starts a new mirror. Two of the failure states are "do not exist" and "degraded". Degraded refers to a disk failure that has been recognized by the system and knows that the disk has not been restored. In this case, VSAN recognizes the "degraded" state and immediately creates a new mirror for the affected object, because since it knows that the disk will not recover soon, it does not have to wait for 60 minutes. The "non-existent" state means that VSAN does not know whether the disk will recover soon, either because the host fails or suddenly moves the disk, and the system starts with a 60-minute timeout.

Q: what happens if SSD fails in a VSAN cluster?

A: SSD is in front of the disk group and acts as a read cache / write buffer. If the SSD fails, the disk group and all its stored components are marked as degraded. Then, if there is enough disk capacity, VSAN creates a new mirrored copy where appropriate. For more information, read http://www.yellow-bricks.com/2013/09/18/vsan-handles-disk-host-failure/

Q: does vSphere support TRIM for SSD?

A: no, TRIM is not currently supported, and TRIM will not be used.

This is the end of the FAQ on how to conduct Virtual SAN/VSAN. I hope the above content can be helpful to you and learn more knowledge. If you think the article is good, you can share it for more people to see.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.