Avid® iNEWS® Server

Enterprise Virtualization Reference

Release 1.0

Last Rev. 12 April 2013

 Contents
1 Executive Summary .......................................................................... 3

2 Environment Configuration: net=A.................................................. 4

3 Environment Configuration: net=AB ............................................... 4

4 Backups and Snapshots................................................................... 5

5 Clones and Templates ...................................................................... 5

6 Scheduled Failover / Migration ........................................................ 6

7 Restrictions ....................................................................................... 6

If You Need Help ................................................................................ 8

1 Executive Summary

The Avid iNEWS Server software is supported in a virtualized environment. A hypervisor

that supports the following minimum guest specifications will support iNEWS Server.
Additional RAM and processing can be added to improve performance.

 2 virtual CPUs
 2 GB RAM
 310 GB virtual hard disk
o 200 MB /boot
o 50 GB /
o 4 GB swap
o 128 GB /rp5
o 128 GB /rp5backup
 2 network interface cards
 Red Hat Enterprise Linux 6 x86_64

Avid iNEWS is supported in either a single- or dual-server configuration. Dual-server

configurations are recommended because there will be no downtime in the event of a
virtualization host failure. A triple-server system is not recommended or necessary; the
virtualized environment provides sufficient redundancy mechanisms with a dual-server
configuration.

A cluster of at least three hosts is required for a dual-server iNEWS system. A two-host
cluster is sufficient for a single-server iNEWS system.

A SAN or virtual SAN is required. Network storage allows multiple hosts access to the same
storage nodes. Multiple SANs are preferred, separating the A and B server virtual hard disks.

Avid has validated a three-host VMware vSphere cluster using vCenter and ESXi 5.1.
Additional details specific to that environment can be obtained from your Avid project
representative.
2 Environment Configuration: net=A

In a single-server, or net=A, configuration, only one NIC is required at the guest level. In the
event of a virtualization host failure, the hypervisor’s high availability logic moves hosting
duties to a surviving member of the cluster and triggers a guest boot. This new host must
have access to the storage node containing the A server’s virtual hard disk.

A net=A system should be configured to automatically run the connect A command

followed by the startup command at boot completion. Starting the iNEWS system
automatically minimizes the downtime experienced during virtualization host failure. Contact
Avid customer service personnel for a supported startup script.

System downtime at host failure is expected to be in the range of 10 to 15 minutes. All

clients and programs connected to the server are terminated when the host fails.

3 Environment Configuration: net=AB

In a dual-server, or net=AB, configuration, additional high availability rules are

recommended. A mirror network is also required.

In the event of a virtualization host failure, the hypervisor’s high availability logic moves
hosting duties to a surviving member of the cluster. High availability rules should be in place
to minimize the likelihood of both iNEWS servers residing on the same host and to guarantee
that the new host has access to the mirror network and the relevant storage node. There
should not be an auto-start script in place for net=AB systems. Standard reconnect
procedures must be followed when a failover event occurs.

The mirror network connects the A server to the B server and provides a consistent, low-
latency connection for iNEWS database mirroring traffic and heartbeats. The mirror network
can be a dedicated switch or simply another VLAN on the current cluster service network.
All hosts in the cluster must have access to the mirror network port group. It is recommended
but not required to configure jumbo frames on the mirror network switch and a 4500 MTU
on the mirror network NICs.
 The system experiences no downtime at host failure. Clients and programs connected to the
failed guest are terminated. Clients and programs connected to the surviving guest continue
work after a 30-second pause while the system detects the disconnect and automatically
reconfigures.

4 Backups and Snapshots

The entire guest VM can be backed up easily in a virtualized environment. Integrate one of
the iNEWS guests into the larger backup process used by the cluster environment. It is
recommended to run periodic dbdumps to file for offsite storage. A gzipped, full dbdump is a
few hundred megabytes to a few gigabytes in size. This file is much more portable during a
disaster recovery scenario than an enterprise SAN backup. There are benefits to both backup
mechanisms, and using them in concert allows for multiple recovery scenarios.

Avid does not recommend snapshotting both A and B in a net=AB system, nor is it
recommended to reconnect backups of an A guest and a B guest. Snapshot the A server only.
If a system needs to be reverted to a snapshot or backup, the preferred method is to revert to
the A snapshot and create a clone of that snapshot to replace the B guest. Then boot both
VMs and reboot B after reconfiguring the network and hostname. The databases are
guaranteed to be identical and can be connected safely. The A and B servers can both be
snapshotted only if both servers first perform the proper command sequence—offline; logout
all; shutdown— to ensure no database writes occur in the time between the snapshot of A and
the snapshot of B. Avid does not recommend this method because it involves system
downtime every time a snapshot is taken.

5 Clones and Templates

Cloning and the use of templates are both supported. Templates and clone sources can be
generated at any stage of the installation and configuration process. It is recommended to
have at least one simple template that is only the operating system and iNEWS pre-install
configuration. This image has the most versatility.
 When deploying a template or clone, ensure that files containing hardware-specific
information are corrected before integration into production, such as ifcfg-eth0, ifcfg-eth1,
and 70-persistent-net.rules.

6 Scheduled Failover / Migration

Live migration is supported at the host and storage level. Migration can be used whenever a
host or datastore must be power cycled or placed into maintenance mode.

It is recommended to perform scheduled migrations during periods of low activity.

A net=AB system will experience a disconnect if the guests lose contact with each other for a
period of time greater than 30 seconds. Migration should fall well below that period of time.
vSphere vMotion, for example, results in a loss of communication closer to five seconds, and
a disconnect does not occur.

7 Restrictions

Serial access and serial message output are not supported. Adding serial support to a
virtualized environment increases the cost and complexity of the configuration. All Linux
consoles are available via the hypervisor UI, mitigating the concerns of monitoring the early-
boot output before NICs are initialized. PuTTY CS can be used with SSH communication to
multiplex commands across both servers. If iNEWS was installed with serial communication
active, disable it by setting /etc/grub.conf to the default (zero) kernel and by removing the
“co” line from /etc/inittab.

Host-only networking for the mirror network is not supported. Host-only networking requires
both guests to reside on the same host, forfeiting the redundancy of a net=AB system during
host failure.

Both diskclear and diskcopy are slower in a virtualized environment accessing a SAN than in
a baremetal environment accessing local storage. Allow sufficient time for diskclear when
planning a new system or a reconnect. Because diskcopy can take place on a live system, so
the performance difference will not affect system availability.

VMware-specific vSphere Fault Tolerance (FT) allows uninterrupted service in the event of a
virtualization host failure. At the time of this writing, FT has not been evaluated for
acceptable performance in a production environment, though the technology has been
verified in sustaining an active net=A iNEWS system during host failure. Sites interested in
implementing FT should allow an evaluation time period to determine if production
performance is acceptable. FT guests are limited to a single vCPU and can experience
throttling if the FT logging (vLockstep) network becomes saturated.
If You Need Help
If you are having trouble using your Avid product:

1. Check the latest information that might have become available after the documentation
was published.
You should always check online for the most up-to-date documentation because the
online version is updated whenever new information becomes available. To view the
online versions, visit the Knowledge Base at www.avid.com/support.
2. Check the documentation that came with your Avid application or your hardware for
maintenance or hardware-related issues.
3. Visit the online Knowledge Base at www.avid.com/support. Online services are
available 24 hours a day, 7 days a week. Search this online Knowledge Base to find
answers, view error messages, to access troubleshooting tips, download updates, and
read or join online message-board discussions.