D
David Hodgson
......that is the question!
I'm putting in a Windows 2003 file/printer server, offline files will
be configured on some of the folders. I need this to be available 24/7.
I can afford downtime but only about 30 minutes max. I'm not that
fussed about peformance.
Now I was going to put in an HP ML350 with redundant Power Supply, Dual
XEON, 1GB RAM, RAID 5 SCSI on an HP 641 Smart Array with about 270GB (3
x 146GB) storage.
Spares = because I have 3 of these servers I can afford to buy 1 CPU, 1
motherboard and 1 Smart Array, these can be used if any of my ML350
servers fail.
Upgradability = can buy another 3 SCSI drives taking capacity upto
730GB.
Cost = £1800 for 270GB, £2550 for 730GB
reason for this server = good reliability, performance and good value
for money.
Partitioning = OS part 1, files part 2
Backup = livestate to restore OS in 20 minutes, veritas to restore
100GB of files in about 6 hours
DFS = could put this in but only the folders without offline files
would work.
Now I'm thinking about a cluster, 2 x (HP DC7600, Single P4, 1GB RAM,
RAID 5 SATA Adaptec card with about 900GB (4 x 300GB) storage.
Spares = none needed
Upgradability = Can change hard drives but these would last 3 years at
current storage usage, I'd rather change the servers after 3 years
before I upgraded the HD's.
Cost = £1700 for 900GB
reason for this server = it's cheap, not bad performance, can hold 4
HD's internally using off the shelf HD's.
Partitioning = OS part 1, files part 2
Backup = livestate to restore OS in 20 minutes, veritas to restore
100GB of files in about 6 hours, although files would still be
available if one server died.
DFS = N/A
Now I have some questions I hope some kind people out there with more
experience than me can help me with...
In your experience is a cluster a good option over a single more
reliably server?
Is there a huge administrative overhead when clustering?
What is disaster recovery like in a cluster, easy, hard?
Would this even work?
Would I need to cluster, what about NLB?
Can I run applications on one server and not the other?
What would you do?
If one server died would the other one still be able to serve users?
I know this is a big request but I really think I can gain more
knowledge from you guys, and it's always a good thing to get the
opinion of other IT professionals.
thanks for reading this far
Dave
I'm putting in a Windows 2003 file/printer server, offline files will
be configured on some of the folders. I need this to be available 24/7.
I can afford downtime but only about 30 minutes max. I'm not that
fussed about peformance.
Now I was going to put in an HP ML350 with redundant Power Supply, Dual
XEON, 1GB RAM, RAID 5 SCSI on an HP 641 Smart Array with about 270GB (3
x 146GB) storage.
Spares = because I have 3 of these servers I can afford to buy 1 CPU, 1
motherboard and 1 Smart Array, these can be used if any of my ML350
servers fail.
Upgradability = can buy another 3 SCSI drives taking capacity upto
730GB.
Cost = £1800 for 270GB, £2550 for 730GB
reason for this server = good reliability, performance and good value
for money.
Partitioning = OS part 1, files part 2
Backup = livestate to restore OS in 20 minutes, veritas to restore
100GB of files in about 6 hours
DFS = could put this in but only the folders without offline files
would work.
Now I'm thinking about a cluster, 2 x (HP DC7600, Single P4, 1GB RAM,
RAID 5 SATA Adaptec card with about 900GB (4 x 300GB) storage.
Spares = none needed
Upgradability = Can change hard drives but these would last 3 years at
current storage usage, I'd rather change the servers after 3 years
before I upgraded the HD's.
Cost = £1700 for 900GB
reason for this server = it's cheap, not bad performance, can hold 4
HD's internally using off the shelf HD's.
Partitioning = OS part 1, files part 2
Backup = livestate to restore OS in 20 minutes, veritas to restore
100GB of files in about 6 hours, although files would still be
available if one server died.
DFS = N/A
Now I have some questions I hope some kind people out there with more
experience than me can help me with...
In your experience is a cluster a good option over a single more
reliably server?
Is there a huge administrative overhead when clustering?
What is disaster recovery like in a cluster, easy, hard?
Would this even work?
Would I need to cluster, what about NLB?
Can I run applications on one server and not the other?
What would you do?
If one server died would the other one still be able to serve users?
I know this is a big request but I really think I can gain more
knowledge from you guys, and it's always a good thing to get the
opinion of other IT professionals.
thanks for reading this far
Dave