VMware vSphere Replication – Bandwith and Overhead Explained (updated)

Bandwith yang dibutuhkan untuk vSphere Replication Technology

Berikut ini adalah cara menghitung Bandwidth untuk vSphere Replication for Teknologi Replikasi dari VMware :

Untuk lebih jelasnya, terdapat VMware Knowledge Based yang membahas mengenai hal ini, yaitu :

Summary dari artikel tersebut adalah :

Contoh kasus, jika ada 500 VMs, rata2 200GB disk (70% used), data change rate 5%-25%, RPO average 4 jam dibutuhkan bandwidth sebesar ~500Mbps. Tetapi jika data change ratenya diubah menjadi 10%-25% bandwidth yg dibutuhkan ~1Gbps, 2x-nya

Di vSphere Replication version 6.0 ada feature baru dengan compression untuk menghemat bandwidth https://blogs.vmware.com/vsphere/2015/03/vr-60-compression.html

For most replication workloads, you will likely see compression ratios of approximately 1.6:1 to 1.8:1.

Performance Overhead untuk vSphere Replication

Pada vSphere Replication ada sedikit performance overhead impact ke VM dan ke ESXi. Impact ke VM seitar 2-6% dan pada ESXi hosts sangat minim, ini juga ada dokumentasinya di FAQ berikut vSphere Replication Frequently Asked Questions: https://communities.vmware.com/docs/DOC-27800

Q: What performance information is available about VMware vSphere Replication?

A: Impact on virtual CPU performance for a virtual machine that is protected by vSphere Replication is approximately 2 to 6 percent. In nearly all cases, this is not an issue because the vast majority of virtual machines are not CPU constrained.

Q: What performance overhead exists on the source hosts?

A: The replication scheduler built into VMware vSphere uses an insignificant amount of CPU and memory to compute the replication schedule of all the virtual machines on the host.

Q: What performance overhead exists on the target hosts?

A: VMware vSphere® Replication virtual appliances receive the replication traffic. They utilize compute and networking resources in a manner similar to any other virtual machine. Additional storage load is placed upon the VMware vSphere hosts as vSphere Replication transfers replicated data to storage using the Network File Copy (NFC) protocol. During a full-sync operation, checksum calculation operations are distributed across multiple hosts to minimize CPU impact on any one host. Several factors influence where contention might occur. For example, if there is much bandwidth – 1Gbps, for example – the amount of NFC traffic might tax the host or even the underlying storage system.


Spesifikasi VM untuk SRM dan vSphere Replication Appliance

Spesifikasi VM yang dibutuhkan untuk SRM, ini terdokumentasi di Installation Guide SRM sebagai berikut

VMware vCenter Site Recovery Manager Documentation: https://www.vmware.com/support/pubs/srm_pubs.html

Site Recovery Manager System Requirements: http://pubs.vmware.com/srm-60/index.jsp#com.vmware.srm.install_config.doc/GUID-384B628F-35C8-4C96-9B36-ACCEBE6C6792.html

  1. SRM – Windows Server 2008/2012, 2 vCPUs, 8GB vRAM, 70GB Disk (5GB Min), 1 GbE NIC/Network
  2. VMware vSphere Replication – 2vCPUs, 4GB of vRAM, 10GB & 2GB disks


Inspired from Bayu’s email. Good Info to share with customers bro.

Be Sociable, Share Tweet about this on TwitterShare on LinkedInShare on FacebookEmail this to someone