Disk Organization Schemes

Since this thread is resolved & m unable to find the create new thread button, I decided to post on a resolved thread, hope u guys won't mind. I have a major issue for which I really need some suggestions. I have to come up with a disk organization scheme for a company & it should reflect areas like storage capacity,performance,user organization,backup practices,redundancy etc, wat on earth is such a scheme I dunno, Google doesn seem to be much help, cud someone enlighten me on this?Secondly, I also want to know how to expand an IAMPD soure code & tarball it contains. My apologies to this thread's starter for latching onto his thread.

Posts merged. - Moderator

I have a major issue for which I really need some suggestions. I have to come up with a disk organization scheme for a company & it should reflect areas like storage capacity,performance,user organization,backup practices,redundancy etc, wat on earth is such a scheme I dunno, Google doesn seem to be much help, cud someone enlighten me on this? Secondly, I also want to know how to expand an IAMPD soure code & tarball it contains. Does anyone have any idea how I can go about this?

These things depend on so much and it really centres around the companies size, existing infrastructure, budget and the type of data involved.

There's no simple answer for this, except you'll need to thoroughly research the above mentioned things and design (on paper) a plan or a "scheme" addressing the issues you mentioned.

Things you might need to consider around each topic (and this is only what I can think of right now):
Storage capacity: Cost vs MB of storage, SAS vs SATA, etc. What kind of RAID implementation should be used (eg: RAID-1, RAID-6, RAID-60 and many more)? How expandable is the storage?

Performance: You'll need to identify potential bottlenecks. Areas to look at include RAM, I/O performance of storage areas and what impact RAID has on that, CPU, network performance (a key one) through the servers adapter/s and the network in general, WAN performance, etc. Also consider operating system performance for the servers particular role if this is also in question. It's quite possible you'd need to conform to the companies existing server OS though.

User Organisation: This depends on the OS. Microsoft is quite particular with best practices for setting up users and permissions and you should just refer to those if you are using Windows Server products.

Backup Practices: You'll need to determine what gets backed up, how frequently and where to. Things to consider here are size of the backups (sometimes limiting where and how to backup), which areas are to be backed up and at what intervals. There are also a number of ways to backup data. You may elect to only backup changes each day, with a full backup at the end of the week (please consider all options carefully). You'll need to have recent backups accessible within short notice, so there may be some on-site backups. There must also be off-site backups. This can be done via 3rd party companies, or online service, or simply giving the boss an external HDD at the end of the week. It really depends on the size and needs of the company.

Redundancy: An important and costly area. This is often what sets server hardware/software apart from your everyday workstation. There are a number of ways to approach it as well. If you're using virtual servers, VMware has a number of (costly) options which decentralise your servers, allowing redundancy across hardware. This allows for boxes to be switched off, upgraded, moved, etc without downtime in the system. Very cool, very expensive. Taking it down a big notch, you've got redundancy within a server box, such as dual PSU so if one fails, you can pull it out and change it while the system runs off the other. Also RAID for hard drives, with the exception of RAID-0. You'll also need to decide which spare parts you have available on hand. Other aspects of hardware redundancy are found in your network. This touches on load balancing as well. Then down to software redundancy, where critical services such as DNS, AD, etc, are hosted by multiple servers. This could also be fine tuned with load balancing.

There's probably more to consider still, perhaps some other folk can fill in the gaps I left.