Intelligent system to prioritize servers for envisaged requests
A workload manager uses on-band and off-band metrics to select a host server in a cluster to handle a connection request. The on-band metrics include CPU usage, memory usage, and vulnerability metrics. The off-band metrics include hardware component error logs. Utilization and vulnerability scores are calculated for each host server from the on-band metrics. A reliability score is calculated for each host server from the off-band metrics. A health score for each host server is calculated from the vulnerability and reliability scores. The health score is used to exclude unhealthy host servers from consideration. A priority score is calculated for each host server from the utilization, vulnerability, and reliability scores. The host server that has not been excluded and has the greatest priority score is selected to handle the connection request.
1. An apparatus comprising:
a load balancer comprising:
memory; and
a workload manager processor that:
collects on-band metrics comprising CPU usage, memory usage, and cumulative common vulnerability scoring system vulnerability metrics from each of a plurality of host servers of a cluster;
collects off-band metrics comprising hardware component error logs from each of the host servers of the cluster;
calculates a utilization score U for each of the host servers based on the CPU usage and the memory usage;
calculates a vulnerability score V for each of the host servers based on the vulnerability metrics;
calculates a reliability score R for each of the host servers based on critical errors indicated in the hardware component error logs;
calculates a priority score P for each of the host servers based on weighted calculated values of U, V, and P;
calculates a health score H for each of the host servers based on calculated values of V and R; and
responsive to receipt of a new connection request:
excludes from consideration ones of the host servers for which H fails to meet a predetermined threshold; and
assigns the connection request to one of the host servers that is not excluded from consideration and has a better score P than other ones of the host servers.
2. The apparatus of claim 1 wherein the workload manager calculates the utilization score U=((100−pCpuUsage)+(100−pMemUsage))/2, where pCpuUsage is percentage of CPU usage and pMemUsage is percentage of memory usage.
3. The apparatus of claim 1 wherein the workload manager calculates the vulnerability score V=1−(cvss_high+cvss_critical)/cvss_cum, where cvss_cum is the cumulative common vulnerability scoring system value and cvss_high and cvss_critical are High and Critical ratings, respectively.
4. The apparatus of claim 1 wherein the workload manager calculates the reliability score
R
=
1
0
0
*
Total
No
.
of
Logs
with
Uniq
(
High
)
and
Uniq
(
Critical
)
Total
No
.
of
Components
-
100
5. The apparatus of claim 1 wherein the workload manager calculates the health score H=(R+V)/2.
6. The apparatus of claim 1 wherein the workload manager calculates the priority score P=(Wtu*U)+(Wtr*R)+(Wtv*V), where Wtu, Wtr, and Wtv are weights.
7. A method comprising:
collecting on-band metrics comprising CPU usage, memory usage, and cumulative common vulnerability scoring system vulnerability metrics from each of a plurality of host servers of a cluster;
collecting off-band metrics comprising hardware component error logs from each of the host servers of the cluster;
calculating a utilization score U for each of the host servers based on the CPU usage and the memory usage;
calculating a vulnerability score V for each of the host servers based on the vulnerability metrics;
calculating a reliability score R for each of the host servers based on critical errors indicated in the hardware component error logs;
calculating a priority score P for each of the host servers based on weighted calculated values of U, V, and P;
calculating a health score H for each of the host servers based on calculated values of V and R; and
responsive to receipt of a new connection request:
excluding from consideration ones of the host servers for which H fails to meet a predetermined threshold; and
assigning the connection request to one of the host servers that is not excluded from consideration and has a better score P than other ones of the host servers.
8. The method of claim 7 comprising calculating the utilization score U=((100−pCpuUsage)+(100−pMemUsage))/2, where pCpuUsage is percentage of CPU usage and pMemUsage is percentage of memory usage.
9. The method of claim 7 comprising calculating the vulnerability score V=1−(cvss_high+cvss_critical)/cvss_cum, where cvss_cum is the cumulative common vulnerability scoring system value and cvss_high and cvss_critical are High and Critical ratings, respectively.
10. The method of claim 7 comprising calculating the reliability score
R
=
1
0
0
*
Total
No
.
of
Logs
with
Uniq
(
High
)
and
Uniq
(
Critical
)
Total
No
.
of
Components
-
100
11. The method of claim 7 comprising calculating the health score H=(R+V)/2.
12. The method of claim 7 comprising calculating the priority score P=(Wtu*U)+(Wtr*R)+(Wtv*V), where Wtu, Wtr, and Wtv are weights.
13. A non-transitory computer-readable storage medium storing instructions that when executed by a computer cause the computer to perform a method for using a computer system to assign a connection request to a selected host server in a cluster of host servers, the method comprising: collecting on-band metrics from each host server of the cluster, wherein the on-band metrics comprise CPU usage, memory usage, and cumulative common vulnerability scoring system vulnerability metrics; collecting off-band metrics from each host server of the cluster, wherein the off-band metrics comprise hardware component error logs; calculating a utilization score U for each of the host servers based on CPU usage and memory usage; calculating a vulnerability score V for each of the host servers based on the vulnerability metrics using the cumulative common vulnerability scoring system; calculating a reliability score R for each of the host servers based on critical errors indicated in the hardware component error logs; calculating a priority score P for each of the host servers based on weighted calculated values of U, V, and P; calculating a health score H for each of the host servers based on calculated values of V and R; and responsive to receipt of a new connection request: excluding from consideration ones of the host servers for which H fails to meet a predetermined threshold; and assigning the connection request to one of the host servers that is not excluded from consideration and has a better score P than other ones of the host servers.
14. The non-transitory computer-readable storage medium of claim 13 wherein the method further comprises:
calculating the utilization score U=((100−pCpuUsage)+(100−pMemUsage))/2, where pCpuUsage is percentage of CPU usage and pMemUsage is percentage of memory usage;
calculating the vulnerability score V=1−(cvss_high+cvss_critical)/cvss_cum, where cvss_cum is the cumulative common vulnerability scoring system value and cvss_high and cvss_critical are High and Critical ratings, respectively;
calculating the reliability score
R
=
1
0
0
*
Total
No
.
of
Logs
with
Uniq
(
High
)
and
Uniq
(
Critical
)
Total
No
.
of
Components
-
100
calculating the health score H=(R+V)/2; and
calculating the priority score P=(Wtu*U)+(Wtr*R)+(Wtv*V), where Wtu, Wtr, and Wtv are weights.