Sound Strategy
Ten Things to Look for
in Your ASP's Data Center
If you use or plan to use an ASP to host software and data critical to your
survival, it's wise to make certain that the ASP's data center operation is
up to the job. Here are ten requirements you ought to look into before taking
the plunge.
- Internet backbone: Is the data center connected to the Internet backbone
or only one step away? The longer the chain, the more potential for delay
and problems with intermediate and uncontrolled servers.
- Redundant connections: Is the data center connected to the Internet via
at least two physically separate adequate bandwidth connections maintained
by two different suppliers? Telephone cables are sometimes accidentally
cut by errant backhoes. You vendor should have backup.
- Backup power supply: Can the data center generate its own electricity
for an extended period of time? It's not enough to have battery backup.
That's important for immediate back up but diesel or gas generators are
needed for longer-term problems.
- Physical security: Does the data center limit entry to the center building,
the computer floor and the cages where the servers operate? Firewalls don't
prevent physical intrusion or interference.
- Redundant data storage: Does the data center maintain RAID 5 or some equivalent
data redundancy system that allows hot-swapping of disk drives? Reloading
from backup tapes is appropriate for physical catastrophe but not normal
day-to-day hardware malfunctions.
- Scalable processing: Does the data center have excess computing capacity
so that it can immediately handle unanticipated processing spikes? Spikes
may be one or more orders of magnitude higher than the normal daily level
of activity.
- Internet security: Does the data center have adequate filtering routers,
redundant firewalls, and other systems to prevent hacking? Thousands of
hackers spend their days looking for ways to make use of or do mischief
to Internet servers. It's a serious problem.
- Clustering technology: Does the data center employ hardware and software
systems so that failure of a server will be transparent to the user? Computers
fail. What happens to the work you have underway when that happens?
- Change management: Does the data center have a rigorously managed change
process? Bringing in new equipment or installing new software can cause
problems. Is testing thorough? Can the center revert quickly to what worked?
- Redundant data centers: Does the vendor host multiple, geographically
separated data centers that mirror one another? Earthquakes, floods, tornados,
hurricanes, fires and other disasters can wipe out a data center. Your vendor
should have at least one spare.
© Copyright 2001 by Sound Internet Strategy. All rights
reserved
