Assignment 6: Difference between high availability and Data

Recovery

High availability and data recovery are often two terms that through around data
center industry for too long. But often people get confused or imagine it to be same,
but in fact it is very different. The very definition of high availability is that
something that is always available. The technological aspect of such term can be said
like something that is protected by various technologies to prevent it from becoming
unavailable. So this somethings can be a server, databases, or even a table inside a
database. High availability is designed to make sure there is no data loss. It also
ensure there is less down time and do high-speed takeovers from one server to
another.
Disaster recovery is more of a pre-planned approach taken to ensure the data or the
components to run an infrastructure is available during the time where normal repair
works cannot fix the issue and have to think about the greater picture with reasonable
time frame limits .The damages caused may include an entire depletion of a data
center, or a complete server depletion etc. Data recovery in total is part of a larger
business survival method. And even in such cases while during a data recovery
procedure we may have to loss some amount of data, since we have to access from
different location sites .We even uses different key matrix elements to calculate the
losses and to know how much of down town may occur etc.
In order to make data available all the time we should know what trigger the problem
in the first place. It may be due to database maintenance or for performing an upgrade
or due to unplanned reason such as server failure or I/O subsystem failure or even
human error. So once we know about the problem the next step would be to plan a
high available strategy. There are two or three attributes we should consider while
planning this strategy

1. Recovery Point Objective: RPO basically explains the maximum allowable

data-loss when a fail over occur. Even though it may be completely technical
we should always talk with the business aspect of the company before making
any decision on RPO. If the business wants zero percentage data loss they may
have to spend huge amount to work out such plans, if not they should
negotiate and find a suitable data loss scenario in which they can both afford
as well as cause less of a problems.

2. Recovery Time Objective: RTO is basically maximum allowable downtime

when different types of failures occur. Assuming like your entire data center is
hit by a tornado we should explain how long it take to recover from such an
event, Is there any other data centers we can use, Is there any new hardware
we can plug in , Is there any off shore back-up sites ,we should consider all
these scenarios and make it available if any such events occur. The business
side should always be aware of this and should always have the budget,
infrastructure requirements etc to meet such scenarios.
Depending upon the availability percentage if it four nine or five nine the downtime
per year may decrease.
Tools used For High availability & Data Recovery

1. Backup and Restore Methods: Keeping backup up of all the files in regular
intervals. This will intern helps u in recovery process if something major
happen. We can also Use Secondary Restore Server were we restore our
backup everyday and keep in the secondary server .So we know our backups
will work and we can make this process completely automated

2. Replication: Replication kind of handle both high availability and DR

scenarios .We will have a source data base or sometimes called as public data
base .A Log reader reads all the process happening in this database and
distribute it towards other distributed database. This distributed database can
then push out this transaction to any number of subscribers that are required.
Since there is no hardware independency between these databases and
subscribers we can make it available from different locations. We can even
give write commands, but may cause lots of confusion regarding queries etc.
Replication method is one of the easiest ways to make sure data is available
when needed.

3. Clustering:
Clustering is mainly used in high availability scenario .In clustering you will
have two-piece of physical hardware. Both the physical servers will be
connected towards the same-shared storage (SAN). So all the database files
from MDF, LDF etc. live on this storage. Using virtual servers if something
happened to any server we can fail over to other server. This failover happens
so continuously that from end users perspective there is no delay or failure.
The nodes communicate with each other using heartbeat technology .if they
didnt get the acknowledgment message based on the conditions and factors
they assume the server had occur a crash and other server take over .SO
basically speaking this technology only help us if one particular server failure
occur
We use ISCSI SAN to perform clustering in case of data recovery
solution.iSCSI works through Ethernet it will have multiple small computing
system interface spread across geographically and data is mirrored in each of
these location.

4. Mirroring
Synchronous mirroring: This technique is used as a high availability
method .In this technique there is no shared storage in synchronous
mirroring, instead there will be different servers with each having internal
memory. Once the application is running it gets connected to a single server,
even though all the transaction, reading/writing all is happening at first
server it is automatically shared with the second server as well .So if the first
server goes down second server kicks in. So we can automate this failover
process by adding a witness server.
 Asynchronous mirroring: This synchronous process is used as disaster
recovery technique. Where there will be two servers and constantly details of
the first server will be shipped towards the second one. The servers may be
placed in two geographical places. They are connected with each other using
network connection and if the network gets busy it will become hard to get
connected.
The disadvantage of both mirroring technique is that we can only failover
one database at a time.

5. Log shipping:

The primary server constantly do data transfer to backup transaction logs share
through network. Other boxes copy these files from backup and restore them
internally towards their own servers.

So these secondary boxes can be placed in various geographical conditions.

We can have one in the same data center other in different data center etc. By
effectively changing the schedules of copying and restoring jobs we can
effectively manage backup without the problems of querying.

6. SAN REPLICATION
The primary server is connected with the SAN storage .The SAN will
have a shared storage where it gets connected to another storage in may be In
different place. Whenever the server does an insert/update etc the primary
SAN tracks whats changed and update it on the secondary SAN. So whenever
a primary server gets fail the secondary SAN will get connected to a backup
server and it will kick in the procedure where it gets stopped. We should
manually change the SAN settings in such a way that it is compactable with
the second server.

7. Virtualization Replication:
In virtualization replication there will be virtual servers, which will run on
many virtual nodes connected in a datacenter. All these nodes save their data
in a SAN storage of the same data center. The SAN storage in turn share all
 this data to a secondary SAN media located in a different location. In the case
of disaster the standby virtual machine can be turned on and the data can be
accessible from the secondary SAN. So during this procedure we may have to
manually re-write the IP subnet number apart from that it is the easiest
migration available.

Conclusion:

Both High availability technique and data recovery methods are integral parts
of data management within a company. Following these procedure not only help
companies cope up with downtime but also make the data available all time even in
the case of a big natural disaster. A company is no good without their user data; an
application that suffers more downtime is no good application at all. So in order to
solve all these problems companies invest large amount of their budget for all these
techniques mentioned above. The write High availability and data recovery procedure
brings a balance between recovery time and recovery cost.

Bibliography

1. http://www.navigatecorp.com/the-difference-between-high-
availability-and-disaster-recovery/
2. http://www.redbooks.ibm.com/redpapers/pdfs/redp4669.pdf
3. http://sqlmag.com/blog/don-t-confuse-high-availability-
disaster-recovery