Today’s application-centric and data-driven economy demands always-on access. Organizations just cannot afford their storage infrastructure to go down when all the mission-critical data is stored and protected there. Being always-on gets challenging as storage hardware and component failures are bound to happen over time, power outages disrupt connectivity, etc.
- Circumvent storage failures and outages
- Enable recovery at a remote secondary/DR site
- Ensure point-in-time recovery of data from last-known good state
The first line of defense is what we are going to focus in this blog, and specifically cover use cases for scaling out in steps to increase high availability. SANsymphony addresses local site problems such as hardware/component failures by replicating data within a site or metro cluster in real time.
2-Node HA Grid | Single Site | 2-Way Mirror
Here we make use of two separate SANsymphony instances (nodes): one as a storage virtualization deployment using disparate array, and the other as a converged storage with HDDs/SSDs built-in and/or directly attached to the SANsymphony node.
Data is copied synchronously between the two physically separate nodes. These nodes can be hosted within a room, a site (can be different rooms, or in different buildings/sites. The distance depends on the latency of their connection. In this example, both deployments are within a site. The virtual disks – A1, B1, C1, and D1 – belonging to the first SANsymphony node are synchronously mirrored and their replicas – A2, B2, C2, and D2 – are available in the second SANsymphony node. When a failure occurs, the application access to the stored data will fail over to the second SANsymphony node. This immediate access to the replicas ensures continuous data access for the applications. End-users and the applications will even not notice the failover because it is a fully transparent process.
It is to be noted that the second SANsymphony node can also host the original dataset and synchronously copy it to the first one, making synchronous mirroring bi-directional.
Such a two-node HA grid provides up to four nines availability (99.99%), while a typical single-node or single storage installation provides only 2 nines availability.
3-Node HA Grid | Single Site | 3-Way Mirror
To increase availability further, a third synchronous copy of the data could be added. In our example, let’s configure a three-way mirror by adding one additional SANsymphony node which is using direct attached JBODs/JBOFs as converged storage deployment.
In this high availability setup, we have created two replicas of the original data. The virtual disks with suffixes ‘2’ and ‘3’ correspond to the two replicas. So, even when one instance is down, there are two remaining synchronous datasets. Thanks to the same failover processes as with a 2-node HA grid, this configuration would tolerate failure in a further instance without end-users and the applications getting affected.
This three-node, three-way mirror deployment further increases HA and business continuity and provides up to six nines availability (99.9999%).
2-Node HA Grid | 2 Sites | 2-Way Mirror
For enhanced protection against larger outages (e.g., fire, power outage, etc.), the SANsymphony grid could be stretched across different buildings, campuses, or sites. As mentioned earlier, the distance of the SANsymphony nodes depend on the latency of their connection. In fast and low latency network environments, this could be up to 100 km. Expanding the two-way mirror deployment across two sites, we now have two SANsymphony nodes – one in each site – with synchronous mirroring set up between them. Here, the virtual disks on the node in Site I are getting mirrored to the node in Site II, providing cross-site resiliency within a stretched metro storage grid.
3-Node HA Grid | 2 Sites | 3-Way Mirror
Consequent to the previous use case, let’s scale out with a third SANsymphony node to create a three-way mirror between two sites.
Here again, we establish a long-distance redundant connection across a stretched metro storage grid; but this configuration increases availability and resiliency with the capability to withstand failure in up to two nodes.
Multi-Node HA Grid | 2 Sites | 2- & 3-Way Mirrors
Next, we are going to scale out with six SANsymphony nodes across two sites with a combination of two-way and three-way mirrors that work cross-sites. In the diagram below, we have examples of some mirrors that have virtual disks with their replicas on the same site, and some having their replicas at the second site.
A key takeaway from this multi-node HA deployment is that you can leverage any SANsymphony node in the grid to store the original dataset, while also storing on it the replicas from any other node in the grid. You can decide in which node you want what dataset or replica. And based on your requirements you can choose to use a two-way mirror or a three-way mirror for each virtual disk.
SANsymphony supports scaling up to 64 nodes in a grid which means there can be any combination of two-way and three-way mirrors for each virtual disk across up to a maximum of 64 nodes spanning two or more sites.
Also take note of the different types of SANsymphony deployment models across the various nodes – storage virtualization of SANs, converged SAN on servers, JBODs/JBOFs. Even HCI deployment of SANsymphony can be factored into the HA grid. And these can all change as required without any negative impact to the applications accessing data.
Load Redistribution Within The HA Grid
Another helpful capability of SANsymphony is to transparently migrate data between the different nodes without any impact on business operation. This helps with optimal resource utilization and customized placement of original data and replicas within the multi-node HA grid. There will be no impact to the applications accessing the data and no downtime.
These are just some examples demonstrating the flexible scaling of SANsymphony to increase data availability and storage resiliency in your environment. Based on your business needs and IT requirements, you can configure high availability with SANsymphony the way that suits you best. Contact DataCore to learn more about achieving high availability and data resiliency using SANsymphony in your environment.