Our application services are backed up at several levels and we also offer our customers options that enable them to run backups on their own equipment hosted on their own premises or, for larger implementations, at one of our co-location facilities.
Operating System Level Snapshots
Each of our production servers are imaged and then backed up automatically, generally scheduled for 1am Sydney Time. Customer server clusters are organised by timezone, which allows us to stagger the backup window timing to occur in the least busy period for each major business timezone for each of our customers.
The backup process on our AWS EBS volumes for our EC2 instances creates a disk level partition snapshot and saves the image of the full partition state to High Speed S3 storage. The image has recovery points set for each day in a 7 day retention cycle.
The snapshot process on our Company Owned and Co-located Hyper-V volumes is done every 15 minutes for 4 hours, then every 4 hours for 24 hours and finally once per day for a 7 day retention period.
SQL Server Level Backups
Our SQL Servers run a nightly database maintenance plan using several best practice steps such as:
- Re-building Indexes
- Updating Database Statistics
- Backing up Transaction Log Data
- Releasing unused Transaction Space on the Database Log File
- .TRN backups are generated each evening at 10pm Sydney time daily with a full .BAK backup run once every 7 days
- Transaction logging is set to FULL on each of our database servers - which allows us to restore to a database point in time
The SQL backups are in turn backed up via the Operating System Level backups previously referred to above.
Customer Run Backups (On Premises)
ValuePRO offers an on customer premises backup option for your file system and database level backups. We supply a free software client (built upon the rsync backup tool provided by Syncrify http://web.synametrics.com/Syncrify.htm). After arranging for firewall rules to be put in place, we give customers with a relatively small ValuePRO implementation (i.e. less than 50 users) to continuously backup their data during a 7pm to 7am backup window.
Customers can then enjoy the peace of mind of knowing that their data is regularly copied incrementally each evening or weekend to equipment at their own premises (or hosted at another cloud service providers premises and equipment).
Customer Run Backups (Co-location)
For much larger ValuePRO implementations (ie greater than 50 users) we recommend co-location of a 1 or 2 RU appropriately sized SAN within one of our data centres. Contact ValuePRO Sales for further information on the pricing of this option depending on your preferred co-location centre.
This solution will be more cost effective and reliable than attempting to run the on-premises backup solution for larger sites as the cost of bandwidth and computing resources required to keep in some cases 10s of millions of files synchronised between physical locations can be uneconomic for larger implementations.
Return to Service Model
Following a service loss incident, having a backup is only one consideration - the time it takes to recover full service and the potential maximum amount of data loss from an incident are the key business considerations.
Our recovery times for a serious incident involving, as a worst case scenario the loss of at least 1 of our 4 mirrored data centre nodes, will be minimal.
As we live mirror the contents of our servers at block level between data centres with only a typical 30 second gap in synchronisation, the maximum theoretical data loss is approximately 30 seconds worth of data. As we essentially maintain a warm live backup of services at alternate daa centres, systems can be configured to failover automatically in minutes and restore full service from a hardware or software level fault without a noticable business interruption.
We don't enable automatic failover on our servers intentionally by design. It is our experience that a fault should first be reviewed and assessed by an experienced engineer so that we can thoroughly understand and investigate the issue prior to bring service up again at a different node.
Over our past 10 years of operation, ValuePRO has maintained excellent uptime in excess of 99.9% business hours availability.