|
|
|
General |
|
|
|
Solution Environment Scope
Details
|
On-prem to on-prem, same hypervisor
On-prem to VMware Cloud on AWS
VMware Cloud on AWS to on-prem
VMware Cloud on AWS to VMware Cloud on AWS
|
On-prem to on-prem, same hypervisor
Zerto IT Resilience Platform (IRP) Standard supports site-to-site replication with the same hypervisor on both sides (VMware to VMware; Hyper-V to Hyper-V), including bi-directional replication.
Cross-hypervisor replication and/or replication to and from Public Cloud platforms requires Zerto IT Resilience Platform (IRP) Enterprise edition.
|
On-prem to on-prem Hyper-V
On-prem to Microsoft Azure
Microsoft Azure to Microsoft Azure
On-prem to on-prem for VMware vSphere or Physical Servers has been announced end-of-life (EOL). Since August 2018 these scenarios cannot be configured anymore in new deployments; only existing deployments are supported until December 31 2020.
|
|
|
Solution Protection Scope
Details
|
Virtual Machines (VMs)
From the outset VMware Site Recovery Manager (SRM) has been designed for protecting virtualized environments only. This means the SRM platform currently cannot be used to protect physical servers for Disaster Recovery purposes.
VMware SRM can be leveraged for protecting Virtual Machines (VMs). Currently VMware SRM is unable to recognize individual containers.
|
Virtual Machines (VMs)
From the outset Zerto IT Resilience Platform (IRP) has been designed for protecting virtualized environments only. This means the Zerto platform currently cannot be used to protect physical servers for Disaster Recovery purposes.
Zerto IRP can be leveraged for protecting Virtual Machines (VMs). Currently Zerto IRP is unable to recognize individual containers.
|
Virtual Machines (VMs)
Physical Servers
Currently Azure Site Recovery (ASR) does not support protection of Docker disks. Also ASR does not support Windows Server 2016 Nano Server.
|
|
|
Evaluation Methods
Details
|
Free Trial (60-days)
Online Lab
Proof-of-Concept (POC)
A 60-day evaluation copy of SRM Standard is freely downloadble from the Product Evaluation Center after registering online. Only the latest GA version of SRM is available for free evaluation. SRM Standard evaluation is not for production environments.
vSphere Replication (VR) is a feature of the VMware vSphere platform. As such, it can be evaluated as part of the 60-day evaluation copy VMware vSphere.
VMware also offers a SRM hosted hands-on lab (HOL-1405) that lets you deploy, configure and manage SRM in a contained environment, after registering online.
|
Free Trial (14-days)
Proof-of-Concept (PoC)
Zerto IT Resilience Platform (IRP) is freely downloadble after registering online and offers full platform support (complete Enterprise feature set) but is time (14 days) restricted. The free trial version of Zerto IRP can be installed on all commodity hardware platforms that meet the hardware requirements.
|
Free Trial (31-days)
Every instance (VM) that is protected with Azure Site Recovery (ASR) is free for the first 31 days.
|
|
|
|
Architecture |
|
|
|
Deployment Architecture
Details
|
Array-based On-premises:
1 Replication adapter (SRA) per site
VR On-premises/Public cloud:
1 Manager (VRAp) per site
1 Replication agent (VRAg) per source host
1 Filter driver per source host
x Replication Servers (VRR)
Array-based On-Premises:
Storage Replication Adapter (SRA) software specific to each storage array that is used with SRM must be installed on the SRM host at both the protected site and the recovery site. SRM supports the use of multiple SRAs. However, virtual disks from a single VM cannot be spanned across multiple storage arrays from different manufacturers.
vSAN does not require a Storage Replication Adapter (SRA) to work with Site Recovery Manager (SRM).
VR On-Premises/Public Cloud:
vSphere Replication (VR) requires 1 Appliance (VRAp) to be deployed at both the source and target sites for management and coordination.
VR also requires an Agent (VRAg) as well as a vSCSI Filter driver to be deployed on each physical server that is part of a cluster hosting protected VMs (source site). The filter driver is used for tapping into the VMs Input/Output (IO) stream.
VR also requires one or more Replication Servers (VRR) to be deployed on the target site for processing and storing the incoming replication data. The vSphere Replication Appliance (VRAp) at the target site had a VRR embedded.
|
On-premises:
1 Manager (ZVM) per site
1 Replication appliance (VRA) per host
1 Kernel module per host
Zerto requires 1 Zerto Virtual Manager (ZVM) to be deployed at both the source and target sites. The ZVM components need to be installed on a (virtual) Windows Server.
Zerto also requires a Virtual Replication Appliance (VRA) as well as a kernel module (provided as vSphere Installation Bundle or VIB) to be deployed on each physical server that is part of a cluster hosting protected VMs (source site) and on each physical server that is used to process and store incoming replication data (target site). The kernel model is used for tapping into the VMs Input/Output (IO) stream.
|
On-premises VMware/Physical Servers:
1 Configuration Server Machine (CSM) per site
x Process Servers
x Master Target Servers
1 Mobility Service per VM
On-premises Microsoft Hyper-V:
1 Site Recovery Provider (SRP) per site
1 Recovery Services Agent (RSA) per host
Azure Public cloud:
None
On-premises VMware vSphere / Physical Servers:
Microsoft recommends running the Configuration Server Machine (CSM) as a VMware VM that can be deployed from a downloaded OVF template. The CSM runs all on-premises ASR components, which include the configuration server, process server, and master target server.
Configuration server: Coordinates communications between on-premises and Azure, and manages data replication.
Process server: Installed by default on the configuration server. It receives replication data, optimizes it (caching, compression, encryption), and sends it to Azure Storage. The process server also performs automatic discovery of on-premises machines (VMs and/or Physical Servers) and installs ASR Mobility Service on VMs that require protection. Additional, separate process servers can be added to handle larger volumes of replication traffic.
Master target server: Installed by default on the CMS. It handles replication data during failback from Azure. Additional, separate master target servers can be added for failback of larger environments.
The Mobility Service needs to be installed on each VM/Physical Server that requires protection. Microsoft recommends allowing automatic installation of the Mobility Service to be performed from the Process Server.
On-premises Microsoft Hyper-V:
The Site Recovery Provider (SRP) is installed on the System Center Virtual Machine Manager (VMM) server where replication is then orchestrated. In addition a Recovery Services Agent (RSA) needs to be installed on each Hyper-V host.
Microsoft Azure public cloud:
|
|
|
Replication Method
Details
|
Array-based: Continuous (synchronous); Snapshot-based (asynchronous)
VR: Changed Block Tracking
Array-based: VMware SRM support both synchronous (continuous) and asynchronous (snapshot-based) storage replication mechanisms. It depends on the storage hardware manufacturer which of these are supported in the vendor-specific Storage Replication Adapter (SRA).
vSphere Replication (VR): VMware developed the vSphere Replication engine. The method vSphere Replication uses to track changes to a VM is very similar to the CBT mechanism that is part of VMware vSphere Storage APIs – Data Protection. However, it is not CBT, preventing interference with other solutions that utilize CBT. VM snapshots are not
used at the source unless Microsoft Volume Shadow Copy Service (VSS) quiescing is enabled when configuring replication.
|
Journal-based (near synchronous)
Zerto proprietary always-on block-level replication technology is a Continuous Data Protection (CDP) solution that uses a journal. As such there is no snapshot mechanism involved.
|
Log-based CBT (VMware/Hyper-V/Physical)
Snapshot-based (Azure IaaS)
VMware vSphere / Physical Servers: The ASR Process Server writes replication logs to a cache storage account in the Azure target region. For every protected disk, log data is replicated to a Managed Disk in Azure. This disk has the prefix of asrseeddisk. It stores the copy of protected disk and all the recovery point snapshots.
Microsoft Hyper-V: The Hyper-V Replica Replication Tracker (HRRT) tracks all changes as Hyper-V replication logs (.hrl). These log files are located in the same folder as the disks. Each disk has an associated .hrl file that is sent to the end-user organizations Azure Storage Account on delta replication. When a log is in transit to Azure, the changes in the primary disk are tracked in another log file, in the same folder.
Azure IaaS: Recovery points are created from snapshots, and stored in accordance with retention settings in the replication policy. ASR takes crash-consistent snapshots of data by default, and app consistent snapshots if you specify a frequency for them.
CBT = Changed Block Tracking
|
|
|
|
Array-based: Continuously (synchronous); Every x minutes/hours (asynchronous)
VR: Every x minutes/hours
|
Every x seconds
If possible, Zerto creates a recovery checkpoint in the Zerto journal every few seconds.
|
VMware/Physical: Continuously
Hyper-V: Every 30/300/900 seconds
Azure IaaS: Every 5 minutes
VMware vSphere / Physical: In the Replication Policy an RPO threshold is configured. Alerts are generated when continuous replication exceeds this limit.
Hyper-V: In a Replication Policy the replication frequency can be configured to 30, 300 (default) and 900 seconds.
Azure IaaS: ASR creates crash-consistent recovery points every five minutes by default. This setting cannot be modified.
|
|
|
Site Connectivity
Details
|
On-premises to on-premises:
Private network, Public internet, VPN
On-premises to VMware Cloud on AWS:
IPSec VPN, VPN over Direct Connect, L2 VPN
|
On-premises to on-premises:
Private network, Public internet, VPN
|
On-premises to on-premises:
Private network, Public internet, VPN
On-premises to Microsoft Azure:
Public internet, ExpressRoute (Microsoft peering)
Azure Site Recovery (ASR) replicates data to an Azure storage account or managed disks, over a public endpoint. Thus protected data can only be replicated over the public internet with ExpressRoute (Microsoft peering). Replication over a site-to-site VPN isnt supported.
When failing back from Azure to VMware vSphere, data from Azure is copied back to the on-premises VM for whoch private access is required. This means for failback a VPN or ExpressRoute is required.
|
|
|
|
Array-based: Solution dependent
VR: Data Compression
Array-based: Deduplication and compression techniques aimed at reducing remote replication traffic is entirely dependent on what the specific storage solution or what a 3rd party WAN acceleration solution can provide.
vSphere Replication (VR): Data compression can be enabled to save network bandwidth between sites. For this VR utilizes the FastLZ compression library. Bandwidth is not monitored, and there is no way to configure bandwidth throttling. The data replicated is not encrypted, enabling 3rd party WAN acceleration solutions to provide some benefit.
|
Data compression
Signature matching
Bandwidth throttling
Data compression: WAN Traffic Compression can be enabled per Virtual protection group (VPG) to save network bandwith in between sites. The VRA automatically adjusts the compression level according to CPU usage, including totally disabling it if needed. Zerto recommends enabling WAN compression. WAN Traffic Compression is enabled by default when replicating to VMware vCloud Director (vCD).
Signature matching: Reduces the amount of data sent across the WAN. During synchronization of the protected site and recovery site for every VM in a VPG, Zerto maintains a map of disk sectors so that if there is a need to resynchronize sites, the map signatures can be used to ensure that only data where changes occurred are passed over the WAN.
Bandwidth Throttling: Bandwidth used for remote replication traffic can be limited. A maximum bandwidth applies to remote replication traffic from the protection site to all peer recovery sites (MB/sec). Throttling can be time-based by specifying time windows using a 24 hour clock.
WAN = Wide Area Network
|
Data Compression (VMware/Physical)
Bandwidth throttling
Data Compression: The Process server acts as a replication gateway. It receives replication data and optimizes it with caching, compression, and encryption.
Compression isnt used when replicating Hyper-V VMs to Azure. For compression a third-party appliance (eg. Riverbed) needs to be used.
Bandwidth Throttling: Bandwidth used for remote replication traffic can be limited. A maximum bandwidth applies to remote replication traffic from the Process Server in protection site to the recovery site (Kbps/Mbps). Minimum bandwidth is 512 Kbps; Maximum bandwidth is 1,023 Mbps. Throttling can be time-based by specifying time windows for Work Hours (24 hour clock) and Work Days.
In addition bandwidth can be throttled for a specifc VM by adjusting the UploadThreadsPerVM in the VMs registry. Default value is 4; Maximum value is 32.
WAN = Wide Area Network
|
|
|
|
Scaling |
|
|
|
Solution Scalability
Details
|
Array-based: 5,000 VMs
VR: 2,000 VMs
|
10,000 VMs
A single Zerto Virtual Manager (ZVM) can protect up to 10,000 on-premises VMs (VMware or Hyper-V). One ZVM can be deployed per vCenter/System Center environment.
|
Unknown
There is no documented maxmimum number of VMs that can be protected by Azure Site Recovery (ASR).
|
|
|
Host/Server Scalability
Details
|
Array-based: N/A
VR: Unknown (VRAg,VFD); 200 VMs (VRR)
The vSphere Replication Agent (VRAg) and the vSCSI Filter Driver that are installed at the source site do not have any documented maximums.
However, each of the vSphere Replication Servers (VRR) installed at the target site can protect a maximum of 200 VMs.
|
1,500 VMs; 96TB (VRA)
A single Zerto Virtual Replication Appliance (VRA) can protect up to 1,500 on-premises VMs (VMware or Hyper-V) and up to 96TB of storage (sum of all protected disks). One VRA can be deployed per virtualization host.
|
85-225 VMs; 2TB Change Rate (Process Server)
NEW
VMware vSphere: A Process Server can be scaled-up by adding CPUs and memory or scaled-out by adding additional Process Servers. There are currently three Process Server sizes:
- 4vCPU, 8GB RAM, 300GB Cache Disk,
|
|
|
Multi-site Support
Details
|
Yes (1:1, N:1)
VMware SRM supports 1:1 and N:1 topologies. VMware SRM does not support 1:N and N:N topologies.
1:1 = site-to-site replication with the same hypervisor on both sides (VMware to VMware). This include bi-directional replication support.
N:1 = VMware SRM supports a maximum of 10 paired remote sites. Concurrent recoveries can be run from multiple sites.
1:N, also known as fan-out, is defined as the same protected VM that can be replicated to multiple recovery sites.
N:1, also known as fan-in, is defined as protected VMs in multiple sites being replicated to a single recovery site.
|
No (1:1 only)
Zerto IT Resilience Platform (IRP) Standard supports site-to-site (1:1) replication with the same hypervisor on both sides (VMware to VMware; Hyper-V to Hyper-V), including bi-directional replication. Zerto IT Resilience Platform (IRP) Standard supports replication of a VM to a single target, either local (same datacenter) or remote (other datacenter).
1:N, N:1 and N:N topologies require Zerto IT Resilience Platform (IRP) Enterprise edition.
1:N, also known as fan-out, is defined as the same protected VM that can be replicated to multiple recovery sites.
N:1, also known as fan-in, is defined as protected VMs in multiple sites being replicated to a single recovery site.
|
Yes (1:1, N:1)
Azure Site Recovery (ASR) supports 1:1 and N:1 topologies. ASR does not support 1:N and N:N topologies.
1:1 = site-to-cloud replication.
N:1 = sites-to-cloud replication.
1:N, also known as fan-out, is defined as the same protected VM that can be replicated to multiple recovery sites.
N:1, also known as fan-in, is defined as protected VMs in multiple sites being replicated to a single recovery site.
|
|
|
Multi-tenancy Support
Details
|
Yes (limited)
VMware SRM supports multi-tenancy in private cloud scenarios.
VMware SRM does not support VMware vCloud Director (vCD).
|
No
Multi-tenancy support requires Zerto IT Resilience Platform (IRP) Enterprise edition.
|
Yes
Azure Site Recovery (ASR) supports VMware vSphere multi-tenant environments for
- tenant subscriptions;
- tenant subscriptions that are created and managed through the Microsoft Cloud Solution Provider (CSP) program.
|
