In Chapter 8, Active-Passive Solution for VxRail, we understood the advantages of the active-passive solution for VxRail. We learned how to use both VMware Site Recovery Manager and vSphere Replication to extend the disaster recovery feature on VxRail appliances across two locations.
In this chapter, we will learn about the advanced solution in the VxRail system. We will learn what a vSAN stretched cluster on VxRail is. This includes the active-active-passive solution with VMware Site Recovery Manager (SRM). We will learn how to plan and design this solution in this chapter, including the deployment of a VxRail vSAN stretched cluster, and how to extend the disaster recovery feature on a VxRail vSAN stretched cluster.
In this chapter, we're going to cover the following main topics:
In this chapter, you'll need to make sure your workstation (laptop) is running on the Windows platform, and that a web browser is installed on your laptop. The latest versions of Firefox, Google Chrome, and Microsoft Internet Explorer 10 or above are all supported. The VxRail software is running version 4.7.300 or above.
For the network requirements of the vSAN stretched cluster on VxRail, the latency or Round-Trip Time (RTT) between the primary and secondary sites hosting virtual machine (VM) objects should not be greater than 5 milliseconds, and the bandwidth between the primary and secondary sites should be a minimum of 10 Gbps or greater. In the vSAN stretched cluster configurations, the latency or RTT between the primary and secondary sites hosting VM objects and the witness node should not be greater than 200 milliseconds.
For the disaster recovery solution, there is a set of VxRail clusters (a minimum of four nodes each) running at both primary and secondary sites in a healthy state, and the VM replication network for vSphere Replication is deployed and ready across the primary and secondary sites. Additional licenses are required for VMware Site Recovery Manager when defining the SRM recovery plan.
The requirements of the SRM appliance configuration are that we must deploy an SRM appliance at each site. The version of the SRM appliance must be the same. The Domain Name System (DNS) name of the vCenter Platform Service Controller (PSC), vCenter Server, and the SRM appliance must be configured with an appropriate DNS and IP address.
As regards the requirements of the vSphere Replication configuration, we must deploy a vSphere replication appliance at each site. vSphere Replication is configured with a dual-core or quad-core CPU, a 13 GB and a 9 GB hard disk, and 8 GB of memory. The DNS name of the vCenter PSC, vCenter Server, and vSphere Replication appliance has been configured to incorporate the DNS with the relevant IP address.
The vSAN stretched cluster on VxRail can deliver an active-active cluster between two geographically separate locations and synchronously replicate data between sites. This feature allows an entire site failure to be tolerated. It extends the concept of fault domains to data center awareness domains. The vSAN stretched cluster must build on between two separate sites (the preferred and secondary site). Each stretched cluster includes two data sites and one witness site. The witness host contains the witness components of the VM objects. The witness host is used as a decision maker that monitors the availability of datastore components when the network connection between the preferred and secondary sites is lost. The witness host can either be a VM or a physical machine.
Now we will have an overview of the logical diagram of the active-active solution for VxRail. In Figure 9.1, we can see that there are three separate sites: they are Preferred Site, Secondary Site, and Witness Site. The VxRail cluster must be deployed across the primary and secondary sites in an active-active configuration (vSAN stretched cluster), with four VxRail nodes deployed at each site. There are two 10 GB top-of-switches for VxRail network connections at the primary and secondary sites. At the witness site, there is a vSAN witness appliance deployed. The VxRail cluster with eight E560 nodes is managed by the vCenter Server Appliance. The VxRail system's VMs are running on this VxRail cluster, that is, the vCenter Server Appliance, vCenter PSC, VxRail Manager, and vRealize Log Insight:
In the next section, we will discuss the requirements for this solution.
The VxRail cluster must be deployed across two separate sites in an active-active configuration. The following table shows the configuration of each VxRail appliance in the vSAN stretched cluster. The witness host must be installed on a witness site that has independent data paths to each data site. The maximum supported configuration of the vSAN stretched cluster is 15+15+1 (30 nodes + 1 witness). The witness host must be running the same version as the VxRail cluster; for example, if the VxRail cluster is running in vSphere 6.7, the witness host must be version 6.7. The stretched cluster supports the Failure Tolerance Method (FTM) of RAID-5/6 and RAID-1. RAID-5/6 must be in the configuration of vSAN All-Flash:
A vSAN stretched cluster uses fault domain (FD) technology to provide redundancy and failure protection across sites. Fault domains provide the core functionality of the vSAN stretched cluster. The supported number of fault domains in a vSAN stretched cluster is three. A stretched cluster requires three fault domains; these are the Preferred Site, the Secondary Site, and a Witness Host. The preferred and secondary sites are the data sites that are configured in a vSAN fault domain. The vSAN witness host can be either a physical host or a virtual appliance. The witness components are stored on the witness host and provide a quorum to prevent a split-brain scenario if the network is lost between the two data sites. This is the third fault domain. The following table shows the minimum number of VxRail nodes that depend on the VxRail software version and vSAN stretched cluster configuration.
Important note
VxRail appliances support vSAN stretched clusters with vSphere Standard edition. A vSAN Enterprise license is required for DRS.
In the following table, PFTT stands for the Primary Level of Failures to Tolerate, and SFTT stands for the Secondary Level of Failures to Tolerate:
Important note
Erasure coding can only be enabled on an All-Flash vSAN cluster.
In the next section, we will discuss the vCenter Server requirements for a vSAN stretched cluster for VxRail.
Starting with VxRail 4.5.200 and above, either the embedded vCenter Server Appliance or an external vCenter Server can be supported for vSAN stretched clusters. If we choose an external vCenter Server for vSAN stretched clusters on VxRail, the following are the external vCenter Server requirements:
In the next section, we will discuss the network requirements for a vSAN stretched cluster on VxRail.
A vSAN stretched cluster in VxRail requires Layer 2 connectivity for the management network between the preferred and secondary sites. Starting with VxRail 4.7.300, Layer 3 is also supported between the two data sites (preferred and secondary). As regards vSAN network connectivity, we can use either Layer 2 or Layer 3. It recommends a minimum of 10 Gbps or greater bandwidth between the preferred and secondary sites. The network connectivity between two data sites and the witness site must be in Layer 3. Figure 9.2 shows a supported network topology in the vSAN stretched cluster. The RTT, roundtrip latency, should not be more than 5 msec (one-way less than 2.5 msec). The data site to the witness RTT should not be more than 200 msec (one-way, less than 100 msec is acceptable). For 10+10+1 configurations, however, less than RTT-100 msec is preferred. For deployments bigger than 10+10+1, an RTT of less than 100 msec is required.
A vSAN stretched cluster includes the management network, vSAN network, vMotion network, and VM network. The management and vSAN networks must be connected to all three sites (the data sites and witness). In Figure 9.2, we can see that the connectivity for management and the vSAN network is in Layer 2 (the same VLAN ID) between the preferred and secondary sites. The network connectivity between the two data sites and the witness site is in Layer 3 (network routing):
In the next section, we will discuss how to deploy the active-active solution for VxRail.
In this section, we will discuss the deployment of an active-active solution for VxRail. This includes the deployment of a vSAN stretched cluster witness and the configuration of a VxRail vSAN stretched cluster. Let's assume that we are deploying a VxRail cluster (with eight E560 nodes) across the preferred site and the secondary site, and that a vSphere cluster is running at the witness site.
We will deploy a vSAN witness appliance in a vSphere cluster at the witness site. The following is the procedure for deploying the vSAN witness appliance:
Important note
The management network on the witness host must be able to reach the vCenter Server Appliance on VxRail.
Now that we've concluded management of the network and the vSAN network for the vSAN witness host, we will discuss the configuration of the VxRail vSAN stretched cluster in the next section.
In this section, we will discuss the configuration of the VxRail vSAN stretched cluster. The following is the procedure for deploying the VxRail vSAN stretched cluster:
Enable SSH on each VxRail node. The SSH service is disabled by default on VxRail. SSH into each VxRail as root (the preferred and secondary sites). Execute the following command:
esxcli network ip route ipv4 add -n <Witness VSAN subnet/24> -g <Local VSAN gateway>
Enable SSH on the witness host. The SSH service is disabled by default on the witness host. SSH into the witness host as root. Execute the following command:
esxcli network ip route ipv4 add -n <VxRail node VSAN subnet/24> -g <Local VSAN gateway>
In the next section, we will discuss an overview of the active-active-passive solution for VxRail.
If we deployed the vSAN stretched cluster on VxRail between the primary and secondary sites, how could the disaster recovery solution for the vSAN stretched cluster be extended? VMware Site Recovery Manager (SRM) also supports extending the site-level protection to other sites. In Figure 9.37, we can see that there are four separate sites: they are Preferred Site, Secondary Site, Witness Site, and Remote Site. The vSAN stretched cluster builds on VxRail across the preferred and secondary sites, and the vSAN witness host is running at the witness site. The VxRail cluster is running at the remote site. The SRM is enabled with vSphere Replication between the vSAN stretched cluster and the VxRail cluster. If the vSAN stretched cluster has a fault, we could execute the SRM recovery plan to recover all the protected VMs on the vSAN stretched cluster in the VxRail cluster at the remote site:
If you want the details of deploying the SRM solution with vSphere Replication on VxRail, you can go to Chapter 8, Active-Passive Solution for VxRail.
In this chapter, we learned the advantages of the active-active solution for VxRail. We learned how to deploy and configure a vSAN stretched cluster on VxRail, using VMware Site Recovery Manager and vSphere Replication to extend the disaster recovery feature on VxRail appliances across three locations. We acquired the skills to deploy a VxRail vSAN stretched cluster and facilitate the disaster recovery solution across three separate locations in this chapter.
In the next chapter, we will learn about the migration methodology for migrating VMs into the VxRail system.
a. VMware Site Recovery Manager
b. vSphere Replication
c. A vSAN stretched cluster
d. Dell EMC RecoverPoint for Virtual Machines
e. All the above
a. Two data sites
b. Three data sites
c. One data site and one witness site
d. Two data sites and one witness site
e. Three data sites and two witness sites
f. None of the above
a. The network latency between two data sites should be more than 5 milliseconds.
b. The network latency between two data sites should not be more than 5 milliseconds.
c. The network latency between two data sites should not be more than 10 milliseconds.
d. The network latency between two data sites should be more than 10 milliseconds.
e. None of the above.
a. 1 GB
b. 10 GB or above
c. 10 GB or above, and network latency should not be more than 5 milliseconds
d. 1 GB and network latency should not be more than 10 milliseconds
e. All the above
a. Only a virtual server
b. Only a physical server
c. Either the virtual server or the physical server
d. OVF format
e. All the above
a. The Dell EMC website
b. Make a request to the Dell EMC support team
c. Make a request to the VMware support team
d. The VMware software website
e. None of the above
a. One virtual network adapter is used for the management network.
b. Two virtual network adapters: one is used for the management network, and the other is used for the VM network.
c. Two virtual network adapters: one is used for the management network, and the other is used for the vSAN network.
d. One virtual network adapter is used for the vSAN network.
e. None of the above
a. vSphere Replication
b. Dell EMC Avamar
c. Dell EMC RecoverPoint for Virtual Machines
d. Site Recovery Manager
e. VMware vRealize Operations Manager
a. Asynchronization
b. Synchronization
c. Asynchronization with RPO set to 0
d. Synchronization with RPO set to 0
e. All the above
a. One vCenter server and one SRM server
b. Two vCenter servers and one SRM server
c. Two vCenter servers, two SRM servers, and one vSphere Replication appliance
d. Two vCenter servers, two SRM servers, and two vSphere Replication appliances
e. None of the above
a. vCenter Server Appliance
b. VxRail Manager
c. vSphere Replication
d. Site Recovery Manager
e. All the above
a. We select Dual site mirroring (stretched cluster) on the Site Disaster Tolerance menu and select 1 failure - RAID-1 (Mirroring) on the Failure to tolerate menu.
b. We select Dual site mirroring (stretched cluster) on the Site Disaster Tolerance menu.
c. We select 1 failure - RAID-1 (Mirroring) on the Failure to tolerate menu.
d. We select None - keep data on Preferred (stretched cluster) on the Site Disaster Tolerance menu and select 1 failure - RAID-1 (Mirroring) on the Failure to tolerate menu.
e. We select None - keep data on Non-preferred (stretched cluster) on the Site Disaster Tolerance menu and select 1 failure - RAID-1 (Mirroring) on the Failure to tolerate menu.
3.145.54.7