Overview
Customer profiles are too important
to the organization primarily because a big part of data is key to the
long-term health of the business. In a typical organization, mission-critical
data center is an often used
information supporting strategic business processes. This can include
accounting, revenue generation, customer service, logistics, and regulatory
compliance. Hindering any access to or recovery of mission-critical data in
these areas could result in damage to organization's reputation, lost sales and
revenues, loss of customers and regulatory or financial penalties. Purpose-built mission
critical technologies matter. There simply is no substitute.
The Mission-Critical Difference
What it makes
technology mission-critical? Committed mission-critical networks - and every
one of the components that are in them - are so designed to withstand the
harshness, inflexibilities and environmental excesses of daily use.
Availability, redundancy, reliability, and security are all the bywords that
make up the mission-critical and this where the difference begins, because all
mission-critical technologies are expected to work when they are needed in most
difficult of times. No ifs and buts! Mission-critical networks always deliver tailored coverage and as
needed in the community, organization’s daily operations, and the unexpected.
And even though mission-critical solutions require substantial investments, the
returns – capacity to handle growing traffic in a disaster, back-up for the
backup systems, coverage where it is required – remain incomparable.
However, if businesses
do not know what data they have and how to classify it, they be likely to value
it as vitally important, in case. When deciding how best to protect data and
applications, the key essential to positive result rests in identifying what's
mission-critical - that is, what the business cannot live devoid of.
Comprehensive site protection by way of
multiple site SAN configuration
Multiple site
SAN feature leverages the main functionality of Network RAID 10 and 10+1,
resulting in two full copies of the information in every volume, and both
copies are guaranteed to be stored on a different node. This is one of the
advantages of Network RAID 10. Owing to the Network RAID 10’s data outline, one
of the clear results of this system is that each node could go off-line at the
very same time. Meaning that half of the HP StoreVirtual cluster could go off-line,
and the associated applications and volumes are still online, without any
intervention from the storage administrator. This degree of availability cannot
be found anywhere else in the storage industry, and even though it is a
powerful and unique feature, its worth is upgraded and developed still more
with the HP StoreVirtual cluster
split throughout the different
locations. This cluster is understandably exemplified as a single pool of
storage, linked by the same networking set of rules as in a single site. In
other words, in a single data center, half of the cluster can be set
differently on another power circuit and the other half on another. The same standard
networking protocols will continue to operate the storage nodes as a single
cluster as they would if all the nodes were in a single power circuit.
After volumes
are configured with Network RAID 10, should power breaker interruptions occurs,
the other half of the cluster will remain online, with a complete duplicate of
the information in the volumes. Moreover, the applications that are relying on
this data nevertheless have access to it, and consequently, the applications
continue to be online. Bringing it through a step farther, say with two data
centers, if half of the cluster is respectively in both data centers, and in
any case an entire data center goes off for some reason or the other, such as
cooling, power or natural disaster, the other half of the cluster will continue
running with a complete duplicate of the information, and no admin intervention
whatsoever is needed allowing the HP StoreVirtual Storage cluster to
continuously serving the data. This is called as multiple site cluster configuration.
With a configured Network RAID 10+1, a multiple site cluster could span over
three data centers, with all copies of the volume at every data center,
provides a premium level of site availability and redundancy.
Unfortunately,
in a multiple site cluster, Network RAID 5 and 6 are not supported.
The HP
StoreVirtual Storage
The fast
adoption of virtualized technologies
as well as the speedy growth in
data necessitates a reliable shared storage system that is always exactly
available. The dependency on shared storage has altered customer high
expectancies about data availability. Though, an uptime of 99.99% (annual
downtime of 52 minutes) used to be the accepted standard for most
organizations. Today, a customer cannot allow lost time across their environs
and a 99.999% (called 5 nines) availability (downtime of 5 minutes annually) is
the new criterion for data availability.
HP for one has
conducted standard quality reviews of every reported instance of data loss and
data unavailability for every storage systems covered by a support agreement.
The quality review is applicable to all HP enterprise-class storage, networking
products, and servers. In the ensuing review process, HP tabulated hours of
data unavailability as reported by customers. Considering this and the number
of systems still under warranty we can approximate field availability. By
definition, field availability is the host or a server availability to
access information that is in the HP StoreVirtual Storage cluster. Whenever the
host or a server is unable to access the data attributable to connectivity
problems or if the data itself is questionable, the storage cluster is deemed unavailable.
For the past two years, HP was able to establish that the HP StoreVirtual Storage
provides 5 nines: 99.999%
availability or much better in the field, once configured in accordance with
best practices.
Solving the
majority of availability challenges can be done by following established HP
StoreVirtual Storage practices. Best practices analyzer (BPA) which is built
into the centralized management console (CMC) can provide the needed guidance
on compliance with recommended best practices. BPA will compare all
configuration rulings with the best practices and emphasizes those
configurations with issues such as incorrect NIC teaming (or the lack of it),
checking data security with Network RAID, appropriate load balancing across
nodes cluster, and much more.
It is strongly
recommended that BPA be referred on a systematic timetable, specifically right
away after creating any alterations to the storage cluster, such as, but not
limited to, adding up or eliminating nodes, making new volumes, or creating
modifications to the networking configuration.
Migration of HP StoreVirtual volume
Peer Motion on
HP StoreVirtual Storage lets a system administrator to move an HP StoreVirtual
volume from cluster to cluster, online, without of having to reconfigure the applications
or hosts. This is simply done by editing the volume properties, selecting the
Advance tab, and selecting a new cluster from the cluster drop-down box. The
blocks making up the volume on the original cluster will start migrating to a
new cluster, and the LeftHand OS will redirect automatically all proxy requests
for blocks to the correct cluster as the data migration is ongoing. When a
migration is completed, the iSCSI shifts to a new cluster from the host are
automatically restored, on the assumption that the new cluster’s IP address was
added to the iSCSI configuration of the hosting server. A Peer Motion typical
to this case could be a volume containing data for an application that has a
growing performance need. If a volume has started out on an SAS MDL cluster, a
storage system administrator may well utilize Peer Motion in moving the volume
to an SAS-based software cluster. Likewise, if the volume is in an SAS cluster,
the storage administrator could possibly choose to add more nodes to the
cluster to provide added performance to the volume or could prefer to move the
volume to an even better performance level such as an SSD-based software
cluster.
To expound it a
little further, the virtualized storage within the HP StoreVirtual cluster will
mean that any rules concerning data that is tied to physical hardware resources
no longer applies. The virtualization, in effect, allows volumes to be moved
dynamically between different physical hardware clusters, and allows for a
cluster swap, further means, its ability to remove existing storage nodes from
a cluster and replace them with new storage nodes, online, with no loss of data
or data availability whatsoever. In this process, data from the removed storage
nodes is moved into the new storage nodes, and all IO is appropriately and
precisely guided to the right node. Upgrading storage nodes to faster, larger
or newer storage nodes does not, at all, require any downtime, producing a
well-defined, clear policy for any future expansion and growth. A fitting
example for this is, a customer may well start off with a cluster of 8 drive
systems, but as the customer adds up applications and workload to these
clusters, these could reach the performance level or capacity of the nodes.
They could easily be migrated to nodes with 12 or more drives to boost capacity
and performance without bringing any applications offline.
Eventually, the
fusion of the amount of data being generated by practically by every
organization coupled with the standard shifts that are being made by the IoT
(Internet of Things) is without a doubt going to categorically move data
standards from the “nice to have” to the “mission-critical” category. If an
industry doesn’t have the standard or is sluggish on developing them, the clock
is unquestionably ticking towards the point of no return.
Upgrading online with Upgrade Advisor
Upgrading
online give storage admins the ability to put into practice the most recent in
firmware and software to the systems without taking the storage cluster along
for maintenance. Other upgrades might also be available in smaller individual
pieces of software or as a group of software known to the community as a patch
set. There can also be instances where enhancements to the HP StoreVirtual
Storage nodes are only accessible via firmware upgrading or through the
LeftHand OS upgrading (it is referred to as the previous versions of SAN/iQ). A number of these enhancements are
available through major version upgrades, such as SAN/iQ 9.0 to SAN/iQ 9.5.
Irrespective of the kind of upgrade, the Upgrade Advisor in the CMC
instinctively tests the present installed software levels on the nodes in the
storage cluster versus the commonly accessible software releases specifically
those that were published by HP, and forewarns the system administrator once an
upgrade is forthcoming. These software upgrades are envisioned to improve the
availability of HP StoreVirtual systems. Moreover, the Upgrade Advisor provides
the possibility to put into effect the software upgrades, and provides a
checklist of dependencies that might need to be remedied in another place in
the environment, such as in guaranteeing that the HP StoreVirtual device-specific
modules (DSMs) Multipath I/O software
is indeed compatible with the
storage nodes upgrades. Thus, it is favorably suggested that any available
upgrades pinpointed by the Upgrade Advisor be appraised and implemented when
possible.
Zero downtime for maintenance
Unlike in other
storage systems, altering or modifying the properties of a volume requires zero
downtime for either the host accessing the volume or the volume itself, giving
storage administrators the leeway it needed to fine-tune the varying changes of
the requirements. Any volume specifically protected by a Network RAID level
higher than 0 can withstand every storage node going offline, whether it is for
maintenance or due to unforeseen situations. As a result of this, maintenance
can be accomplished whenever you like it in a live environment, without
bringing down applications or hosts. Just simply, select a storage node to
perform the maintenance on, and carry out the maintenance works. Even if a
specific node turns out to be unobtainable due to a software installation that
necessitates a power switch off, or may be a maintenance reboot, the volumes
that are protected with Network RAID higher than 0 would still be available. In
this scenario, a storage maintenance window is no longer a requirement, because
by simply carrying out maintenance on one node at a time, Network RAID will
allow HP StoreVirtual Storage to just proceed in serving the data.
Subsequently,
if there is a requirement for volumes to remain online when one or more nodes
in an HP StoreVirtual cluster or a management group necessitate maintenance,
every volume have to be secured by a Network RAID level higher than 0. Clearly,
Network RAID 10 is the best there is, that is to say, recommended Network RAID
level since this one offers an excellent combination of availability and
performance the system needs. After ensuring that all volumes are secured by
Network RAID, simply continue with the maintenance, one node at a time. After
finishing the maintenance for that node, it comes back online and the data is
synchronized. After the synchronization is complete, the next node in the
cluster can go through the same maintenance work.
The virtualized
storage allows administrators to alter or modify practically every single
attribute of an HP StoreVirtual volume, Network RAID, including size, whether
the volume is provisioned thinly, and all that. Since the fundamental physical
storage is virtualized, there are no solid rules as to where to store the data
or how to configure the volumes. While a good deal of these jobs is just
accounting changes, meaning counting of how many blocks are allocated to a
specific volume, they can be finished promptly with no impact whatsoever on
system performance.