JetStream DR Administrator's Guide (version 5.x)

Concept

Description

JetStream DR Software ("JSDR")

JetStream DR is a software application that enables Disaster Recovery capabilities for virtual machines and their data. JSDR can be obtained and installed to protect on-premises environments. A version known as "JSDR for AVS" is available from Azure Marketplace to protect workloads in the Microsoft Azure cloud.

Protected Domain
(“Domain”)

A Protected Domain contains VMs specified by the user that should be protected and restored together. All VMs of a Protected Domain are replicated to the same container on the storage site.

Storage Site / Object Store

The storage site is the environment where the object store maintains continuously updated storage objects containing Domains. The storage site can be located at the recovery site, or it can be in a different location.

Primary Site

The original environment where VMs to be protected initially run. When a Domain is created, the current site is appointed as the "primary site."

Recovery Site

The destination environment where Domains from the object store will be rehydrated to and subsequently run.

Protected Site

The protected site is the "owner" of the site where VMs of the Domain are currently running. The protected site can be associated with the primary site or the recovery site. Protected site ownership will change during the course of DR operations.

Restore

The disaster recovery function when the local site (where the user is currently logged in) becomes appointed the "primary site" of the Domain and recovered VMs continue to run from there.

Failover

The disaster recovery function when the local site becomes appointed the "recovery site" of the Domain. It is OK to continue running the Domain at the recovery site without failing back to the primary site.

Failback

The process where Protected Domains are returned from a recovery site back to the primary site. Ownership and operation are are shifted from the recovery site back to the primary site.

Continuous Failover
("CFO")

Continuous Failover is a mode of operation where the protected and recovery sites remain synchronized during normal operation to minimize the amount time necessary to complete DR recovery.

Point-in-Time Recovery ("PITR")

Point-in-Time Recovery allows VMs and data from a user-specified time in the past (last "known good" point) to be recovered, tested, and then restored. Protected Domains can be constructed with PITR capability allowing their VMs and data to be restored to a previous point in time within a window of protection specified by the user.

Runbook

A set of instructions that automate failover and failback operations of a Protected Domain to control VM startup sequence and timing, and VM configuration parameters.

Management Server Appliance (“MSA”)

The MSA is a virtual appliance that runs as a plug-in to vCenter Server. It collects and maintains statistics relevant to the protection of the VMs in the cluster(s) managed through vCenter. It also provides administrative functions, such as selecting VMs for protection, etc. The MSA can be conveniently managed from the vSphere Web Client, and its functions can also be directly accessed via CLI or RESTful APIs.

IOFilter

An IOFilter is a virtual machine component that runs in ESXi and intercepts all IO operations between a VM and its corresponding virtual disk(s). IOFilters are used to seamlessly examine currently existing disk data and replicate new data written to virtual disks without impacting system performance.

Protection Modes

Data of protected VMs can be written to primary storage and the replication log using two methods: (1) “Write-through”: The IOFilter acknowledges completion of a write operation back to a protected VM only after it has received acknowledgement of the write from both the replication log and the primary storage. (2) “Write-back”: The IOFilter acknowledges completion of the write operation to the protected VM upon receiving acknowledgement from the replication log volume only; the write to primary storage is asynchronous.

DR Virtual Appliance
("DRVA")

IOFilter captures data for replication but does not communicate directly with the object store. The DRVA is a virtual appliance that maintains the replication log store and manages the transfer of VMs and their data to the object store. The DRVA also manages functions such as in-line compression and garbage collection. 

Each cluster must contain at least one DRVA to manage it. Additional DRVAs can be added to a cluster for enhanced performance, but only one DRVA can be assigned per host.

Recovery from Object Cloud Virtual Appliance ("RocVA")

A virtual appliance that is temporarily deployed for recovery, failover, and failback operations. During recovery, the RocVA facilitates rehydration of VMs and their data from the object store. RocVAs are automatically removed from the system when no longer needed.

Representation VM
("RVM")

During failover, an RVM is temporarily created for each VM being rehydrated to assist with the recovery process. RVMs are automatically removed from the system when no longer needed.

Replication Log ("RL")

The replication log is the record of all data replicated to the object store. Each Protected Domain uses a single replication log for all the VMs belonging it.

Replication Log Volume

The replication log volume is a shared, non-volatile memory resource reserved for JetStream DR software. It is exposed to DRVAs as an iSCSI LUN or VMDK-based virtual disk. The replication log volume is used to maintain replication logs, garbage collection metadata and other metadata.

Background Replication

This is the process of reading existing data from a virtual disk at the protected site and copying it to the object store. Background replication traffic is heaviest when protection is initiated because all contents of protected virtual disks must be copied before protection can start.

Foreground Replication

This is the process of continuously identifying newly generated data at the protected site and copying it to the object store. This is an on-going process and only occurs as new data is written to protected virtual disks.

Write Throttling

Write throttling is the temporary disabling of data replication to the object store. Write throttling may be applied when network bandwidth is unable to keep up with background or foreground replication traffic. In cases where throttling parameters are exceeded, the system will switch to a "bitmap" mode to absorb a network traffic spike until backlogged data can be replicated to the object store. Write throttling is automatically managed by the system. If it is detected that write throttling is periodically being applied, it may indicate issues with the network that should be investigated and resolved to ensure optimal performance.

Garbage Collection

When invalidated data is no longer needed at the protected site it is removed from the object store by a "garbage collection" process. This minimizes unnecessary consumption of storage space at the object storage site.

Ownership ("IO Fencing")

To maintain integrity of a Protected Domain, only one JetStream DR site at a time can write to (update) the object store for a Domain. Multiple sites can connect to the Domain to read its contents and status, but only the "owning" site DRVA can write to the Domain's container in the object store. The DRVA of the owning site updates the Domain status to maintain its ownership and the write lock ("fencing out" DRVAs of other sites from writing to the object store).

Ownership of the Protected Domain can shift between the local protected site or the remote recovery site depending upon the action currently being performed.