Cisco Cisco IP Contact Center Release 4.6.1 Release Note
Cisco ICM Software Release 6.0(0) Bill of Materials
Ethernet adapters has been proven to introduce packet delivery/reception problems capable of generating latency sufficient to cause
problems with ICM function. Also, it is common in ICM components for an Ethernet adapter to have more than one IP address,
which is often not supported in redundant/load balancing configurations. Some other unsupported options include specific RAID
configurations and the hardware/software configurations that go with them. This is discussed further in the “RAID Controllers,
Arrays, and Storage Options” below. Redundancy in other components (power supplies and the like), however, is not only
supported but recommended.
problems with ICM function. Also, it is common in ICM components for an Ethernet adapter to have more than one IP address,
which is often not supported in redundant/load balancing configurations. Some other unsupported options include specific RAID
configurations and the hardware/software configurations that go with them. This is discussed further in the “RAID Controllers,
Arrays, and Storage Options” below. Redundancy in other components (power supplies and the like), however, is not only
supported but recommended.
SCSI/ATA/IDE Hard Disk Drives
SCSI Hard Disk Drives are recommended for components that are expected to have high levels of disk I/O. This includes, but is not
limited to, servers running database applications. Components where ATA/IDE drive use is acceptable are noted as such. Ultra
3/Ultra 160/Ultra 320 SCSI (based on availability) comes highly recommended. “High-end” installations can also benefit from higher
rotational speed (e.g. 15K rpm for SCSI) drives, which further reduce access latency. (10K rpm drives are typically standard options
for SCSI; ATA/IDE drives are typically 7200 rpm.)
limited to, servers running database applications. Components where ATA/IDE drive use is acceptable are noted as such. Ultra
3/Ultra 160/Ultra 320 SCSI (based on availability) comes highly recommended. “High-end” installations can also benefit from higher
rotational speed (e.g. 15K rpm for SCSI) drives, which further reduce access latency. (10K rpm drives are typically standard options
for SCSI; ATA/IDE drives are typically 7200 rpm.)
RAID Controllers, Arrays, and Storage Options
Properly configured RAID disks are recommended for most components, due in part to the ICM/IPCC application being deemed
“mission critical”. There are, however, many considerations to take into account with RAID. First, “software” RAID (without an
embedded or add-on hardware controller) is generally not recommended, as its only advantage is some level of redundancy at the cost
of decreased overall I/O performance. In general, all types of hardware RAID controllers are acceptable (embedded and add-on),
with significant factors being the number of channels and on-controller cache of the controller. There are varying recommendations
around configurations, but several primary factors to keep in mind.
“mission critical”. There are, however, many considerations to take into account with RAID. First, “software” RAID (without an
embedded or add-on hardware controller) is generally not recommended, as its only advantage is some level of redundancy at the cost
of decreased overall I/O performance. In general, all types of hardware RAID controllers are acceptable (embedded and add-on),
with significant factors being the number of channels and on-controller cache of the controller. There are varying recommendations
around configurations, but several primary factors to keep in mind.
In general, the performance of an array increases with the number of physical drives contained within. The following are
representative configurations for a sample 144GB disk (listed in order of potential performance, from highest to lowest): 8 x 36.4 GB
RAID 10 (1+0), 4 x 72.8 GB RAID 10 (1+0), 2 x 145.6 GB RAID 1. Maximizing the number of physical drives also increases the
overall fault tolerance of the array. The limiting factor would be the number of physical drive bays available in the platform.
representative configurations for a sample 144GB disk (listed in order of potential performance, from highest to lowest): 8 x 36.4 GB
RAID 10 (1+0), 4 x 72.8 GB RAID 10 (1+0), 2 x 145.6 GB RAID 1. Maximizing the number of physical drives also increases the
overall fault tolerance of the array. The limiting factor would be the number of physical drive bays available in the platform.
Multi-channel controllers offer a true performance advantage only when connected to discrete drive bays (“cages”) or backplanes. In
situations where there are multiple drives connected to a single SCSI backplane connection, splitting those drives into multiple arrays
and across multiple channels on the attached controller provides only a small amount of physical separation and negligible
performance advantage. Ultimately, all communication is funneled through the single physical backplane connection – and that
remains the bottleneck, regardless of the capabilities of the controller. Multiple controller channels can be of significant advantage
when there are multiple drive “cages” and backplane connections. In this case, each channel of the controller can connect to a
separate backplane connection, and arrays split between the channels and backplanes can take advantage of the increased throughput
as well as increased resiliency.
situations where there are multiple drives connected to a single SCSI backplane connection, splitting those drives into multiple arrays
and across multiple channels on the attached controller provides only a small amount of physical separation and negligible
performance advantage. Ultimately, all communication is funneled through the single physical backplane connection – and that
remains the bottleneck, regardless of the capabilities of the controller. Multiple controller channels can be of significant advantage
when there are multiple drive “cages” and backplane connections. In this case, each channel of the controller can connect to a
separate backplane connection, and arrays split between the channels and backplanes can take advantage of the increased throughput
as well as increased resiliency.
There are several RAID configurations which can be deployed. Some manufacturers use proprietary configurations. Consult the
manufacturer’s documentation to determine what “standard” RAID configuration their proprietary platform is based on.
manufacturer’s documentation to determine what “standard” RAID configuration their proprietary platform is based on.
In general, mirroring or RAID 1 is recommended as adequate for Contact Center server applications. 32MB of on-controller cache is
a good baseline for PGs, Distributor AWs, and ICM Routers. 64MB should be considered the starting point for a Logger or AW with
HDS and/or WebView option. Higher-end Loggers, HDSs and/or WebView machines should move to 128MB or more cache and
consider a dual or multi-channel controller if additional throughput or number of arrays is needed.
a good baseline for PGs, Distributor AWs, and ICM Routers. 64MB should be considered the starting point for a Logger or AW with
HDS and/or WebView option. Higher-end Loggers, HDSs and/or WebView machines should move to 128MB or more cache and
consider a dual or multi-channel controller if additional throughput or number of arrays is needed.
Note: Given this recommendation, and the fact that the MCS servers as currently specified in this document utilize a single
channel RAID controller with only 128 MB Battery-Backed Write Cache, these servers will not be appropriate for higher-
end (e.g. carrier class) deployments.
channel RAID controller with only 128 MB Battery-Backed Write Cache, these servers will not be appropriate for higher-
end (e.g. carrier class) deployments.
Higher-end (Large Enterprise) deployment of ICM or IPCC should consider RAID 10 (1+0). With the increase in capability and
cache size of even an “entry-level” embedded RAID controller, the move to RAID 10(1+0) from a basic RAID 5 array is often just
the cost of 1 additional HDD. Cache, in this configuration, should be a minimum of 256 MB. Dual or multi-channel controllers are
recommended to maximize throughput and offer the option of spreading the array across multiple backplanes or drive “cages” for
cache size of even an “entry-level” embedded RAID controller, the move to RAID 10(1+0) from a basic RAID 5 array is often just
the cost of 1 additional HDD. Cache, in this configuration, should be a minimum of 256 MB. Dual or multi-channel controllers are
recommended to maximize throughput and offer the option of spreading the array across multiple backplanes or drive “cages” for
3