HP StorageWorks Enterprise Virtual Array Online Firmware Upgrade Best Practices (5697-6388, January 2007)

3 Cus tomer testimonials about
online u pgra des
A number of EVA customers have successfully used the online rmware upgrade feature. The following
are a few of these successes.
Customer 1: A large telecommunications company
Customer 1 is a very large EVA customer with hundreds of storage systems. They use Windows, HPUX
and Solaris hosts. Applications include MicroSoft Exchange, MicroSoft Sequel Server, Oracle, and a
number of custom applications.
They decided
that they needed to upgrade nearly 100 EVA5000 storage systems from VCS 3.020 and
3.025 to VCS
3.028. The magnitude of this upgrade made scheduling of ofine upgrades complicated
and painful
. The customer’s ma nag em ent team requested that online upgrades b e used to reduce the
down time o
f systems d uring the upgrade process.
The HP customer support team star ted worked with engineering to create guidelines and processes that
would make this effort successful. Many of those guidelines are included in this document.
The team monitored the storage systems to ndappropriatetimestodotheonlinecontrollerrmware
upgrade. They found that even systems that were supposedly fully utilized generally had times where
workloads were at a level where the online upgrades could be done successfully. Once an upgrade
window was identied, the upgrade was scheduled. At the tim e of the upgrade, storage system
throughput was monitored to validate that the workload was within the acceptable guidelines. If the
demand on the storage system was at an acceptable level, they proceeded with the upgrade.
With nearly 10 0 storage systems to upgrade there were a few issues. The most signicant issue was
servers o ccasionally losing connectivity to the storage system. The customer carefully and methodically
inv est
igated the p roblem. Their investigation revealed that a third-party multitasking driver had been
congu
red with paths that no longer existed. Consequently, when the storage system was resyncing, the
drive
r did a failover to path that was nonexistent. In subsequent upgrades, verication of failover paths
was added to the preparation process and the p roblem was eliminated.
On several storage systems, a time period could not b e found where the work load fell off to the desired
levels. In these cases, t wo d ifferent approaches were used to upgrade those storage systems.
Sche
dule an ofine upgrade.
Identify the host(s) generating the h igh workloads, and then shutdown just those hosts to reduce
the workload to an acceptable level.
Customer 2: A large software company
In the rst half of 2006, customer 2 performed at least six online controller rmware upgrades in their
data center. Five EVAs were used for internal ap plications and one EVA was used to host Microsoft
Exchange data. All the upgrades were successful.
The most interesting case is the upgrade of the Microsoft Exchange storage system. After careful analysis
of the EVA to evaluate the dem and on the storage system, the controller rmware upgrade was done
around 4 PM during a business day.
Customer 2 also successfully upgraded two EVAs from 2C6D to 2C12D while host I/Os are running.
HP StorageWorks Enterprise Virtual Array online rmware upgrade best practices
25