The advent of solid state disk (SSD) has the potential to be a game changer to the world of enterprise storage. The main advantage of SSD-based storage is faster performance and reduced latency, as well as lower heat output compared to traditional disk.
Organisations needing the highest levels of speed and execution are turning to SSD for exactly these benefits, but the top tier applications they store on SSD are usually the ones in greatest need of protection against disaster as well.
This therefore makes solid business continuity planning essential. Integrating SSD into your storage can affect existing disaster recovery plans, and costs associated with them, if you are not careful in your approach.
Companies traditionally tackle business continuity through duplication, protecting their IT assets and applications through keeping a copy of all data. At the base level, enterprise storage is typically deployed with one block of storage at the primary site to support the production data and applications and then a second block at the replication site. This can be a second company location or hosted remotely by a service provider. Activity on the primary storage is replicated over to the secondary location in order to provide a comprehensive backup.
While effective, this approach can be expensive, as you have to buy double the amount of identical hardware and potentially double the volume of software licenses. If an investment has been made a certain brand of storage for the production site, then the same brand and model can be required at the secondary site as well.
This approach is normally linked to synchronous replication, where an action at the primary site is reproduced at the secondary site immediately. This limits the distance that can exist between the two sites while keeping the data sets synchronised, as light can only travel at a certain speed over the network. For linking locations that are further away geographically, or where network speeds are not high, then an asynchronous approach would have to be used. This ensures that data is kept in the same order, but it does not have to be done at the same time on both sites.
For companies introducing SSD arrays into their networks, there are decisions to be made when it comes to their business continuity strategy. Three options present themselves:
- Implement multiple SSD arrays at both the primary and secondary site (full duplication)
This is the more expensive option as it relies on the traditional approach of duplicating everything but it ensures full data and application protection. In a disaster situation, you would have a back up of your primary array on SSD with all the data. However, this would have to be balanced against how likely this level of storage would be required, the need for SSD performance at the DR site and the cost to implement.
- Implement multiple arrays at the primary site only
This offers high availability to protect against hardware failure at the primary array; however, this does not protect against full site failure events. Another data protection method would be required for disaster recovery.
- Go down a route where replication is decoupled from the storage array
This enables the freedom to choose the storage technology best suited for your needs and much greater flexibility in relation to failover capabilities to the disaster recovery site.
One obvious benefit to moving the replication away from the storage array is the cost. Instead of investing in multiple duplicated arrays for continuity, other storage resources can be used at the secondary site. It is also possible to mix replication methods to give greater availability and disaster recovery. For example, combining two SSD-based arrays at the primary site can ensure availability should one storage platform fail, and complementing this with data replication over to the secondary site for DR.
Mixing it up
Many companies are evaluating SSD-based storage as part of their server virtualisation implementations. Typically, customers are running a mix of physical and virtual server environments, while some organisations operate a combination of hypervisors such as VMware and Microsoft Hyper-V. This means that replication between platforms must be taken into consideration, as well as replication between different storage platforms.
As businesses grow their IT infrastructures, any upgrade or new storage install in the production site does not necessarily have to lead to an automatic upgrade or new storage project at the replication site. Instead, SSD arrays can be used for performance at the primary site while other storage platforms can be used at the recovery location. This can help reduce cost, as organisations may not require the same level of performance from their secondary systems.
Taking this approach to storage hardware can also mitigate the need for investment in new resources when a move becomes necessary. Rather than implementing all-new storage systems across both primary and secondary sites, the storage that was previously used for production can be moved over to the DR site. By using replication at the OS level, data and applications can be replicated separately from the underlying storage layer.
The main thing to remember is that data can be moved in real time to the location where it is needed, wherever that location might be or whatever platform it might run on. Replicating in real time across from the primary to the secondary site provides up to date data for continuity and recovery.
Protecting data is essential to business. Potential gaps in data replication not only damage the business bottom line but can also have serious compliance and governance implications. At the same time, implementing new and innovative technology that can improve overall business performance means that continuity planning has to be managed as an on-going process, rather than being a one-off activity.
The temptation to explore the advantages of SSD-based storage cannot be ignored. Yet, in tough economic times, companies will not necessarily want to invest in SSD throughout their IT infrastructure; they may not be able to afford multiple arrays, or might not see the economic benefits from this approach. By looking at alternative technologies for data replication, companies can get the advantages of having SSD-based storage, while not being tied into expensive storage level replication. This alternative enables enterprise IT to have their SSD storage cake and eat all the benefits of robust business continuity as well.