• Post Reply
  • Bookmark Topic Watch Topic
  • New Topic

How to ensure High Availability in EJB centric Architecture

 
Miles Bells
Greenhorn
Posts: 5
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
Hi,
I'm designing a low traffic EJB centric system, one of my requirements is High Availability (99.99%)

To meet the requirement I plan to use:
- 2 load balancers (session affinity)
- 2 web servers
- 2 application servers
- 2 clustered DB

Do you think this is enough?
Are there any other design consideration to take care in your opinion?

Thank you for your feedback
Miles
 
K. Tsang
Bartender
Posts: 3583
16
Android Java
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
Is this for the JEA assignment?

Other things to consider is data synchronization/replication, hot/cold server standby, network bandwidth and IO throughput (eg for data sync)

These may not directly relate to HA but more application performance during peak hours (many users). Because if say syncing (large) data happens every 5 mins may affect user experience during peak hours, which to the extreme may cripple the application hence become not available.
 
Miles Bells
Greenhorn
Posts: 5
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
No, I'm still studying for the JEA 6, Part I

Availability for me is still unclear, apart from put load balancers and to mirror all the servers I don't know how to improve it.
And the availability of the DB is also a concern for me, some DB supports the cluster, is this the only way to increase their availability?


Thanks for the help
Miles
 
Jeanne Boyarsky
author & internet detective
Marshal
Posts: 35279
384
Eclipse IDE Java VI Editor
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
Miles,
Are the two servers sitting right next to each other? This could be a problem is something happens to the physical site.
 
Miles Bells
Greenhorn
Posts: 5
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
Jeanne Boyarsky wrote:Miles,
Are the two servers sitting right next to each other? This could be a problem is something happens to the physical site.



Good point

After reading this article: http://en.wikipedia.org/wiki/Availability
and checked the mean time to failure (MTTF) of the server disks (about 1.600.000 hours) I wonder why I need to mirror the servers? This MTTF alone can archive a 99.99% on a single machine...
 
Jeanne Boyarsky
author & internet detective
Marshal
Posts: 35279
384
Eclipse IDE Java VI Editor
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
Miles,
Good question. A few reasons:
1) That's an average time to fail. Some disks fail sooner. Some fail MUCH sooner.
2) Once the server fails, you need to switch somewhere. There isn't enough time to buy a new server and still meet five 9's.
3) Other things can happen to a computer besides the hard disk failing.
 
K. Tsang
Bartender
Posts: 3583
16
Android Java
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
Talking hard disk failing, choosing what storage approach is also a factor (eg mirror, RAID5 etc).
 
  • Post Reply
  • Bookmark Topic Watch Topic
  • New Topic