Why do we need Database HA Software like SQL Always On, Oracle Dataguard etc

Question

I am learning the very basics of High Availability, SAN etc. and hence this question may sound stupid to experts, but would greatly help me if you answer it. Let’s say I am using an Enterprise SAN setup. I understand that any database e.g. SQL stores the data in a File which is stored on the SAN. Now if let’s say I enable array-based replication to another array may be in another Data center, then my database file will continuously be replicated in the second data center. Whenever the first data center is lost, I can use the replicated file in second data center to bring up the data and database. Then what exactly is the role played by various HA solutions like SQL Always On, Oracle Dataguard etc.? Thanks a ton in advance to people who rely.

Answer 1

Mission Critical systems are just that. Mission Critical.

We test this, and plan for it to happen but dependency upon redundancy is no real redundancy. Are both installation yours? Are third parties involved? For how many days (or hours) can you bear the extra latency?

Andrew, Katrina, Sandy. If it is not just you, how high are you on the priority list for your third parties? They may be blue chip long on promise, but short on delivery when spread thin on the ground.

When you do recover, there will be exceptions, dead letters, and possibly a period before you can confirm eventual consistency. Functionally you may be fine, but your brand may be damaged, your stock price devalued.

Ultimately it is meeting a requirement that is in turn driven by a real need or by risk mitigation strategies. I'm sure it is sold to people who do not need it. But for the people that do, 100% uptime is an absolute, not something one rounds to for the Annual Report. These products are actively marketed to the 'C' suite for these reasons.

Answer 2

When you build IT solutions, some requirements to consider are how resilient your IT solution needs to be based on the business process criticality. there are two aspects of the solution that you need to consider.

1. High availability (HA): Which determines how resilient you application needs to be and it is usually expressed in 9s (nines) for example 99.99% availability is called four nines availability which is the equivalent of about one hour of unplanned downtime. HA in Oracle this is accomplished usually by using oracle real application cluster (RAC), which gives you availability even in the event when a node in the cluster goes down. in SQL Server this would be SQL Server availability group.

2. Disaster recovery(DR): determines methods and technologies to provide business continuity in case of disaster in other words when your application High Availability (HA) features are no longer responding to requests. in Oracle this would be a dataguard replicating your database from one cluster in one datacenter to a second cluster at the remote location. SQL server also provides similiar features such as log shipping, alwas-on, and availability groups. disaster recovery capabilities can be measured by two metrics: Recovery time objective (RTO) which determines how long it would take for the backup site to be fully funtional in case of loss of the primary data center. and the Recovery Time objective (RPO) which determines business data loss tolerance.