Table of Contents
Advertisement
Many applications can be adapted to take advantage of MSCS' failover feature. These applications must
be able to use TCP/IP and use remote storage for its data. (Applications that keep a significant state
information in memory are not the best applications for clustering because any data not stored on disk is
lost at failover.) Applications that can be failed over can support MSCS API or not. Those applications
that support the MSCS API are MSCS-aware. MSCS-aware applications can be registered with the
Cluster Service to receive status and notification information and use the MSCS API to administer
clusters.
Although MSCS protects the availability of data, not the data itself in a cluster, the Windows NT Server
has built-in features including mirroring and redundant array of inexpensive disks (RAID Level 5,
stripping with parity). The following table lists common points of failure in a Windows NT Server
environment and suggests protection methods:
Point of Failure
Network hub
Power utility
Disk
Server hardware (CPU or memory)
Storage
Server software (operating system
or applications)
WAN links (routers and dedicated
lines)
Dial-up connection
Computer
Any client that can connect to Windows NT Server through TCP/IP will work with MSCS. This includes
Microsoft MS-DOS®, Microsoft Windows® 3.x, Windows 95, Windows NT, Apple® Macintosh®, and
UNIX®. MSCS does not require any special software on the client for transparent recovery of services
that connect to clients through standard IP protocols.
For more information on MSCS, see http://www.microsoft.com/ntserver/support/clusterfaq.asp#basics.
DIGITAL Server Cluster Kits for Windows NT Configuration Guide
MSCS
Redundant networks
Failover
Failover
Failover
-
Other
Consider stocking replacement
network components
Uninterruptable power supply (UPS)
provides power for 5 to 20 minutes for
a proper shutdown
RAID array
DIGITAL Servers 5000 and 7000
series provide redundant power and
cooling
StorageWorks RA3000, 450, 7000
offer redundant power, cooling,
cache, controllers, cache batteries
Consider stocking replacement disks,
controllers, SCSI components
Redundant links over the WAN,
providing secondary access to remote
connections
Multiple modems
Multiple clients for the same level of
access
9
Advertisement
Table of Contents
