Approaches to Disaster Recovery in Schools Maurice Cummins, Monte and Ian Ralph, SCEGGS.

Approaches to Disaster Recovery in Schools

Maurice Cummins, MonteandIan Ralph, SCEGGS

This session:

• 3 Parts:– Some time to complete our survey– SCEGGS’ DR journey so far– Review of the survey results

• Aims– Broaden understanding of DR– Source the wisdom of crowds

10 Minutes to complete Survey

• http://www.surveymonkey.com/s/KZ5J9L3

Some Scenarios & Consequences

• Scenarios– A/C failure– Earthquake– Fire– Plane Crash– Disease outbreak

• Consequences– Devices fail– Inability to access site– Key staff gone/unavailable

Where we would like to be

• Initially (1 day or less)– Website– Communications (parent portal e-mail, SMS, LMS)– School Admin system online (contacts, payroll etc)

• After a few days – 1 week (users have devices)– Files– Remote access– Other systems – e.g. Mahara

Where SCEGGS is now

• Tapes Offsite• Purchase bare metal• Rebuild• Restore• Reconnect• Weeks of labour-intensive effort

New technologies – new opportunities

Enabling Technologies

VMs Cloud based VMs SANs SAN to SAN replication iSCSi External Bandwidth – 1Gbps –

10Gbps VOIP/SIP trunking for VOIP

redundancy

Virtual Apps and or Virtual Desktops for key users in DR

Cloud Storage Virtual Networking – software

defined networking (e.g virtual firewalls

Cluster capable servers for redundancy

Considerations for all options

• Monitoring – is your DR solution online?• Access to competent 3rd parties to advise and

minimise risk• Make sure that 3rd party knows your system

before accepting a proposal• Test regularly – you only know if your backup

works when you try a restore• Think DR for future purchases

Options – On Site• Physically remote server room• Replicate Hardware• AC + UPS• Redundant diverse path connectivity• FC tape backup in remote location – doing now

Pluses Minuses

Least Cost High speed connectivity Easily accessible

Vulnerable Connectivity issues if site

destroyed Access issues cf. Christchurch Power issues

On Site physical separation

Tape backup

Diverse Connectivity

Remote server space

Off-site – rack space at other school

• Assumes – Rack space + AC + UPS + High Speed connectivity

• Sufficiently remote• Replicate hardware• Reciprocate

Pluses Minuses Low ongoing costs Safe location(?)

Connection setup complex Not School’s core business Shared connection or purchase

independently (adds to cost) Physical/data security

How much rack space?

Off-site – rack space commercial facility

• E.g. Global Switch $1695/Month for 40RU• Physically remote server room• Replicate Hardware Pluses Minuses

Physically Secure High availability power and cooling Core business Good/redundant connectivity Low ongoing costs Safe location(?)

Connectivity could be complex Access for contractors could be

tricky Need to keep 2 sets of hardware up

to date

Off-site – 3rd Party replication• Integrator like Netstrategy/Sommerville/ASI• Physically remote server room• Replicate our Hardware• Partner certified on your hardware/software

Pluses Minuses

End to end solution Complexity for integrator to sort Physically Secure High availability power and cooling Core business Good/redundant connectivity Specify annual DR test Integrator staff/documentation to

step in if needed

Cost – but expense not cap-ex

Off-Site – Cloud

• E.g AWS• Redundant Data Centres

Pluses Minuses

Low costs until system used Physically Secure High availability power and cooling Core business Good/redundant connectivity Data replicated

Complexity of setup related to addressing/interoperability

No native support of vSphere Partner definitely required Costs grow depending on data

usage/storage/VM Specs Complexity of address space Potential issues of data sovereignty

Off-site – cloud for some services

• Google Docs• Office 365

Pluses Minuses

Relatively inexpensive/FTE option Simple well understood Physically secure High availability Core business Good/redundant connectivity Data replicated Applications also cloud based Generous storage allowance

Limited scope of solution Risk 3rd party ceases to operate Potential issues of data sovereignty Some interoperability issues Complexity for more than simple

workloads Latency when data centre offshore If LDAP used what happens when

AD disappears?

Hybrid• E.G. Cloud (Google docs/Office 365) + On Site

Pluses Minuses

Best of both worlds Lowers requirements for

H/W on site Connectivity Cost

In theory - no interruption to Cloud services

Same issues with Cloud as already mentioned

Same issues with On Site as already mentioned

How to choose?

• Resilience – will the solution survive a catastrophe?

• Recovery Time – how quickly can we be back online

• Complexity• Connectivity• Flexibility• Support• Cost

A selection Matrix

• Weighted scores• Each school needs to develop weights

Survey Data Results Access

http://tinyurl.com/plab65z

Password: Contact Ian Ralph if you require access to this resource.

Thanks

• Maurice Cummins – Monte• James Boyle – Netstrategy• Neil Jackson/John Hildebrant/Ken Yap – AWS• Richard Warren/John Collias – Computelec• Leah Rix – Dell