Method for estimating the availability of an operating server farm

• US 6,496,948 B1
• Filed: 11/19/1999
• Issued: 12/17/2002
• Est. Priority Date: 11/19/1999
• Status: Expired due to Term

• Current Assignee: Unisys Corporation
• Original Assignee: Unisys Corporation
• Inventors: Smorodinsky, Lev
• Primary Examiner(s): Iqbal, Nadeem
• Application Number:
  US09/443,926
• Time in Patent Office: 1,124 Days
• Field of Search: 714/37, 714/38, 714/39, 714/4, 714/6, 714/25, 714/43, 714/44, 714/56, 709/220, 709/221
• US Class Current: 714/37
• CPC Class Codes: G06F 11/008 Reliability or availability analysis
• CPC Class Codes: G06F 2209/503 Resource availability
• CPC Class Codes: G06F 9/505 considering the load

Abstract

A method and an estimator program for estimating the server farm size and the availability of the server farm for a given redundancy factor and a given particular number of clients.

Claims

1. An estimator program that performs method steps for estimating the availability of an operating server farm designed to serve a particular number of clients “

n”

comprising the steps of;

(a) inputting a group of parameters involving at least one server farm parameter and one single-server parameter;

wherein said step (a) of inputting said group of parameters includes the steps of;

(a1) selecting for input said particular number of clients “

n”

for utilizing said server farm;

(a2) selecting for input a maximum single server workload of users;

(a3) selecting for input a normal single server workload of users as a percent of the maximum (peak) single server workload of users;

(a4) selecting for input a mean time to repair (MTTR) a single server;

(a5) selecting for input a mean time to failure (MTTF) for a single server;

(b) generating an estimate of farm size for said server farm;

(c) establishing the server farm reserved number of servers;

(d) providing an estimate of the mean time to failure (MTTF) for said server farm;

(e) estimating the availability of said server farm.

2. The method of claim 1 wherein step (b) for generating an estimate of server farm size includes the steps of:

(b1) calculating the single server normal workload of users as the nearest integer that is not greater than the product of percentage of the single server normal workload of users and said selected maximum single server workload of users divided by 100;

(b2) calculating the estimated server farm size as the nearest integer that is not greater than the ratio of said particular number of clients “

n”

over said normal single server workload of users;

(b3) incrementing said estimated server farm size by one if the remainder of the division of said particular number of clients “

n”

by said maximum single server workload of users is greater than zero.

3. The method of claim 1 wherein step (c) for establishing the server farm reserved number of servers, includes the steps of:

(c1) estimating the server farm peak number of users as the product of said maximum single server workload of users and said estimated server farm size;

(c2) estimating the server farm normal number of users as the product of said normal single server workload and said estimated server farm size;

(c3) estimating the server farm reserved number of users by subtracting said estimated server farm normal workload number of users from said estimated server farm peak number of users;

(c4) estimating on a trial basis the estimated server farm reserved number of servers as the nearest integer that is not greater than the ratio of said estimated server farm reserved number of users over said maximum (peak) single server workload of users;

(c5) verifying said estimated server farm reserved number of servers.

4. The method of claim 3 wherein step (c5) includes the steps of:

(c5a) calculating a trial minimum number of servers which can serve a particular number “

n”

of clients by subtracting said estimated trial server farm reserved number of servers, increased by one, from said estimated server farm size;

(c5b) calculating an estimated maximum number of clients who can be served by said trial minimum number of servers by multiplying said trial minimum number of servers by said maximum (peak) single server workload of users;

(c5c) establishing the said trial server farm reserved number of servers equal to said trial server farm reserved number of servers in the case of a successful verification on condition that said trial maximum number of clients is greater than or equal to said particular number of clients “

n”

;

(c5d) establishing the trial server farm reserved number of servers equal to said trial server farm reserved number of servers increased by one for a case of unsuccessful verification on the condition that said estimated maximum number of clients is less than said particular number of clients “

n”

.

5. The method of claim 1 wherein step (d) for providing an estimate of meantime to failure (MTTF) for said server farm includes the steps of:

(d1) utilizing a Markov algorithm to estimate MTTF while involving (i) the said minimum number of servers which still permit said server farm to serve said particular number of clients “

n”

in the case that said server farm reserved number of servers are inadequate to service “

n”

clients (failure);

(ii) said server farm size;

(iii) the reciprocal of said selected MTTF for a single server;

(iv) the reciprocal of said selected MTTR for a single server.

6. The method of claim 1 wherein step (e) for estimating the availability of said server farm includes the step of:

(e1) subtracting the ratio of said selected MTTR for a single server to said estimated MTTF for said server farm, from one.

7. The method of claim 1 which includes the step of:

(f) estimating the downtime of said server farm.

8. The method of claim 7 wherein step (f) for estimating the downtime per year, in hours, of said server farm includes the step of:

(f1) multiplying the ratio of said selected MTTR for a single server over said estimated MTTF for said server farm, with the number of hours per year (8760).