
Reference Architecture for Active System 1000 with VMware vSphere
Page 23
Performance: Dell Compellent SC8000, with the dual-controller configuration, 8 Gb Fibre Channel
interconnects provides high bandwidth for data flows. This bandwidth is complemented with a large
variety of drives in multiple speeds and sizes. The SC8000 controllers also use virtual port IQNs and
WWNs, thereby enabling higher throughput and fault tolerance.
Drive Types and Automated Tiered Storage: In the Active System 1000v solution, the number of
storage enclosures and the drives in the enclosures can be customized based on customer
requirements. Administrators can mix SSD and SAS drives in the same system, as well as SAS drives with
the same form factor (but different speeds and capacities) in the same storage enclosure. A maximum
of 16 enclosures is supported in Active System 1000v.
High speed drives are assigned to higher tiers and low speed drives to lower tiers. Compellent Storage
Center automatically configures RAID levels for these tiers and automatically moves the data between
the tiers based on access patterns. Compellent Fluid Data storage dynamically moves data to the
optimal tier based on actual usage. The most active blocks reside on high-performance SSD, or SAS
drives, while infrequently accessed data migrates to lower-cost, high-capacity SAS or SATA drives. For
more details, refer to the Automated Tiered Storage web page. Automated Tiered Storage requires
Data Progress licenses for Compellent.
RAID Array Design: Dell Compellent SC8000 supports RAID 5, 6, and 10. The Compellent Storage Center
will dynamically set up RAID based upon the demands of applications accessing data on the storage
tier(s).
Multipath Configuration: In the solution, VMware Native Multipath Plug-In (NMP) is used to provide
multi-pathing. Path Selection Plug-Ins (PSPs) run with the VMware NMP and is responsible for choosing a
physical path for I/O requests. Round Robin (VMW_PSP_RR) path selection algorithm is the
recommended configuration for Compellent Storage Center. Round Robin uses a path selection
algorithm that rotates through all available active optimal paths enabling load balancing across the
paths. This ensures all the paths are used to provide the maximum bandwidth and balance I/O across
the paths/fabrics.
7 Management Infrastructure
Two PowerEdge R620 servers and one Dell Networking S55 1Gb Ethernet switch are used for
management infrastructure. The PowerEdge R620 servers are connected to the Dell Networking S4810
switches using Broadcom 57810 Dual Port 10Gb Network Adapters. The servers are connected to the
Compellent storage through the Brocade 6510 switches using a QLogic QLA2562 8Gbps Fibre Channel
Card.
Note that the Compellent storage is shared between management cluster and compute cluster. The
Compellent storage must be sized so that sufficient bandwidth is allocated for both the management
VMs and compute VMs.
The PowerEdge R620 servers run VMware ESXi 5.1 Update 1 hypervisor and are a part of the unique
vSphere Cluster. VMware High Availability is enabled in that cluster to provide HA for virtual machines.
Admission control is disabled in the VMware HA Cluster. If admission control is enabled, VMware HA
would prevent putting one of the management servers in maintenance mode, since this would violate
HA policy of having more than one active server in the cluster.
Commentaires sur ces manuels