smaller reset larger        English         

Main Menu

All times are in GMT -8 (DST) :: The time is now 3:26 pm.

Sub Menu

Article Data
Article Ref
Written By
Thom O'Connor
Date Created
Fri, 19th Oct 2007
Updated By
Thom O'Connor
Date Modified
Fri, 19th Oct 2007

   Hardware purchasing


What type of hardware should I buy for my CommuniGate Pro deployment?


First, please recognize that this is a very difficult question to answer, in a general way. Clearly, you should buy the best hardware you can buy at the most affordable cost and which satisfies the peak workload requirements without wasting either system resources or money.

That said, we can suggest the following:

1. Use hardware and operating systems that your IT/administrative team is comfortable with - if you cannot administer your environment, it does not matter how good that hardware or OS might be.

2. For hardware purchasing, CommuniGate Systems often recommends “horizontal scaling” using more, smaller systems rather than fewer, larger (more expensive) systems. Using horizontal scaling generally provides the optimal cost-per-message while simultaneously building more redundancy into the system. As a general guideline, CommuniGate Systems

• 1-to-4 (dual) CPU systems or blades, 2-4GB RAM

• Dual-fans/dual-power supplies are generally recommended for system resiliency to common hardware problems.

• For highest levels of redundancy, mirrored internal disks, 36-72GB in size.

• Again, for redundancy, dual-NIC cards trunked/bonded to the network is a good idea, though not required.

• Frontend servers have e-mail (SMTP) queues - these queues can use a significant amount of disk I/O, so at times adding an additional internal disk to these systems or mounting these queues on NAS/SAN/DAS is useful for performance and redundancy, but not always required.

• Backend servers are attached to the NAS/SAN to access a Shared Logical Volume(s) with a Shared File System (NFS or Cluster File System).

• The NAS/SAN device can be sized initially for short-term growth; however, we strongly recommend that future growth is considered when purchasing storage. The NAS/SAN device should have additional disk capacity and must be capable of growing or extending the available logical volumes (or at the least, be capable of adding additional logical volumes available to all Backends). If those logical volumes can be expanded while the filesystem is mounted and the cluster is online, all the better - the Dynamic Cluster should never need to be brought down then in order to expand the storage volume of the environment.

* In general, CommuniGate also recommends for performance benefits that all volume concatenation or striping (RAID) occurs at the array hardware or array integrated-management level, and not through system or software-level volume management.

How Useful Was This Article?      (Rating: 68%    Votes: 5)  

Select a Rating

Article Comments 

There are currently no comments.