Hitachi Virtual Storage Platform (VSP) Architecture
The Hitachi VSP architecture is a modular based enterprise level storage system.
A fully configured VSP will contain maximum of 6 racks with 2 DKC units and 16 DKU units. It can be configured as a single chassis or dual chassis. Each chassis has at least one DKC and one or two DKUs (DKC and DKU are explained later in this post). An illustrated diagram of fully configured VSP is show below.
The main components in VSP are disk controller (DKC) and disk unit (DKU).
Disk controller (DKC) is the heart of VSP. All operations in VSP are handled by DKC.
Disk unit (DKU) is collection of physical hard drives.
Two DKCs are interconnected via grid switches (GSW).
The main components in DKC are
- Frontend directors (FED)
- Backend directors (BED)
- Virtual storage directors (VSD)
- Cache adapters
- Service processor
- Grid Switch
- Power supplies, cooling fans etc.
and DKU contains
- Hard drives
- SAS switch
- Cooling fans
- Power supplies
Details description of components is here
Below diagrams displays front and back sides DKC
DKC design is modular based and most of the components are replaceable units. Virtual directors and cache directors are placed in front side of DKC. FEB, BED, service processor and grid switches/express switches are place in back side of DKC.
Each LDEV is assigned one VSD (virtual storage director) to process I/Os. Each LDEV owned by one VSD at a time. This LDEV ownership is passes temporarily if VSD board fails.
Architecture of VSP
The basic system architecture is shown in the following diagram.
In VSP FEDs, BEDs, cache directors (CMA) and Virtual storage directors are communicated through Grid switches (GSW). All director components are directly connected Grid switches which form Histar-E network.
Channel adapters (FEDs) are for Host connectivity and Disk adapters (BEDs) are for backend disk connectivity.
VSDs are I/O processing boards and responsible for processing I/O from FEDs to BEDs.
Whenever a host sends an I/O request, FED receives the request and communicates with VSD. VSD processes request and sends the information to cache and BEDs. If requested data is available in cache it directly responses with data. If requested data is available in cache it fetches the data from Backend disks and serves the data.
VSP has spare drives in case of any hard drive failure. It supports dynamic sparing. It means if a hard drive is suspected to fail or getting I/O errors more than a threshold value, Storage system starts copying data to spare drive even if it not failed.
In this case data copied to spare drive not recreated. It is a pre-emptive copy.
If drive failed and pre-emptive copy does not exist. Spare drive becomes part of that RAID group and starts rebuilding data by using parity. It is called correction copy.
It is a software which is installed on Service processor (SVP). It keeps monitoring VSP at all the time. Collects the hardware status and error data. It sends collected data to Hitachi Data Systems support center for analyzing. If any hardware occurs Hi-Track monitor immediately reports to service center.
Hi-Track monitor enables HDS support team to access VSP remotely as they were at site.
Hi-Track SVP agent is a tool installed on Service processor (VSP). It performs error monitoring and sends email alerts to customer.