|
Finding the Right Size
Sizing remote access is an ongoing process. Initially, you can estimate the number of users, usage patterns, the amount of data that will be transferred and a host of other variables. Finding the balance between adequate QoS (quality of service), measured by port availability and cost of ownership is difficult. Although remote-access vendors and users want nearly perfect availability, you want efficient use of remote access to reduce cost while providing decent QoS. Balancing those ends is the goal of finding the right size.
First, look at your user population and determine if you need to meet differing levels of service. The IS department in the research hospital wants high availability for key network administrators to access the network remotely.
Certain administrators also need above-average availability as well. The bulk of the remaining user population has similar needs. On-call network administrators may need user-to-port ratios as low as 2:1 or 3:1. Hospital administrators may need 4:1 or 5:1, while the rest of the population can be settled at 8:1 or 10:1.
Segmenting users cuts into available ports. And to effectively provide differing QoS, you must restrict user access to specific ports--taking them out of the available pool. Consequently, segmenting requires more ports than if the ratio were flat (see "Port Requirements" at right). You could reduce the overall port count to 24 or 23, a single T1 or T1/PRI respectively. When all users are at 8:1, the effect is negligible; however, when users are segmented, the ratio for the general population rises to 11:1. Although that may be acceptable, it means more users will be contending for fewer resources, and as the user population increases, you'll have to move fast to keep up with the growth.
Sizing for Growth
Sizing an existing remote-access solution is in some ways easier and increases the accuracy of the resulting predictions, provided you have solid historical data about usage patterns and calling patterns. Logs from remote-access servers provide information about individual and group usage patterns. This is a good indication of when users are accessing services and how long they need access. Records from PBXes or the telephone company can indicate the number of calls attempted--both successful calls and blocked calls (a blocked call is met with a busy signal).
Remote-access logs provide a picture of who uses remote access and when they need service. Actual usage may be very different than expected usage, and monitoring provides the information necessary to identify trends over time and adjust accordingly. For example, if a large percentage of users in a group are online for less than 30 minutes a session, you can adjust the user-to-port ratio a few points without significantly effec
ting QoS. Adjusting the ratio to meet user needs lets you squeeze the most out of your available remote-access ports.
On the other hand, tracking the number of calls attempted (offered traffic per hour), you will have solid data on when users need remote access. If you look at only the remote-access logs and notice that you have two hours daily at 95 percent capacity (this takes into account modem recycle time and training time, for example), you might assume that you need to add ports. However, if the offered traffic indicates a low number of blocked calls, then your setup is probably fine as is.
Proper planning requires you to size for the points of highest utilization, or the average of the highest utilization, not for the average usage utilization. If you size for low or medium utilization, users will have a more difficult time gaining access during peak periods. Of course, it's always a balance between cost and service.
Dividing the number of blocked calls by the total number of calls attempted
results in a measure of QoS. If 13 out of 250 calls attempted are blocked, then 5 percent of all calls get hit with a busy signal. Raise the number of blocked calls to 50, and 20 percent of all calls are blocked. For the general user population, higher rates of busy signals might be acceptable, but for mission-critical applications and users, you'll want to maximize availability.
General Guidelines
The following guidelines will help ensure greater success in installing and managing a successful remote access solution.
· Look for remote-access solutions that are easily scalable.
· Keep thorough logs and review them for trends and sizing.
· Set reasonable access policies and be prepared to adjust them as needed.
· Examine access logs looking for usage patterns for the general population and groups of users.
Mike Fratto can be reached at mfratto@nwc.com.We would like to thank the Renex Corporation, Shiva Corp. and Phil Williams of Children's Hospital Care Center
for their assistance with this article.
|
REPORTS
ANALYTICS
Follow 
