Bartholomew Baskin, OP, October 2008
peter@webcentre.co.nz
mka Peter Hyde, Director, Webcentre Ltd
interim Treasurer, SCA NZ Inc.
All figures in this proposal are expressed in Australian Dollars unless otherwise stated.
Lochac's Internet infrastructure is very good but is currently completely dependent on the support and decisions of key volunteers and/or their employers. Little is done to support these people or to provide any coordination to improve outcomes for them, the Kingdom, SCAA or SCANZ.
Master Del (David Elson) has provided and funded the main sca.org.au server for years, but now wishes to hand on that role. This will involve at least a modest and unavoidable hardware cost -- perhaps $1000 -- and some limited ongoing costs for domain names, SSL secure certificate and the like -- perhaps $200 a year.
A hosting and administration offer from Lord Raoul from St Florians (David Bussenschutt) is on the table for sca.org.au. If this is accepted, hosting will be free, provided bandwidth demands do not affect either the server's operation or his employers' hosting node. This should be avoidable but, should problems arise, the best alternative technically is a US-based host -- adequate solutions range in cost from about $1000 to $1500 a year.
This proposal recommends
The key benefits of this proposal are that it:
Because we will remain reliant on volunteers and donated hosting, there will always be a degree of unpredictability in this area. However, the proposed role and regular funding will make it easier to plan for and support reliable operations, and lessen the chances of a sudden failure of any kind.
At present Lochac enjoys a very good Internet and communications infrastructure, which is essential for a Kingdom of this size and nature. For many groups, their websites and email lists are crucial parts of their recruiting and ongoing operations. Offices, guilds and orders also derive notable benefits, as do the SCAA and SCANZ.
This infrastructure depends on the informal support and contributions (both volunteer labour and costs) of many skilled individuals -- some directly, others through the support and good offices of their employers. The most significant contribution they make is their time, but there are also real dollar costs involved for many of them.
Among those currently providing significant hosting support in Lochac are:
There are doubtless several others who provide some limited hosting, but the first two are crucial. Each of them could, in principle, assume the load of the other and indeed that of almost all the other service providers. However, to do so would be to increase the chances of both burnout and of a serious single-point service failure affecting the whole Kingdom.
Therefore, our best strategy in a volunteer-led environment must be to maintain at least two strong Kingdom-level providers, and encourage them to provide mutual support and redundancy to each other where possible.
Where costs can be clearly identified, e.g. domain name or secure certificate registrations, small items of replacement hardware, etc: fund this directly.
Where costs are being shared with other bodies (employers, private use, other community groups they support): offer an annual ex-gratia contribution towards those costs, commensurate with the benefit to the Kingdom -- even if it might not meet the costs entirely. In many cases, the offer might be declined, but the principle behind it will be appreciated.
Create a Kingdom-level role -- included in an existing officer job description or as a deputy position -- intended to help the Kingdom coordinate, plan and part-fund its infrastructure, while recognising that the core work and hosting will continue to be supported by volunteers for as long as possible.
An outline job description for this role is in Appendix I. If this proposal were substantially accepted by all three funding bodies, I would be willing to take on this role on an interim basis until it can be properly advertised in Pegasus.
SCAA obtains most of its annual funding from subscription fees paid by just over 700 members. SCA NZ receives fees from just over 100 members, plus a significant amount in event memberships (non-member surcharges). Kingdom is funded by the Kingdom levy, paid at events in both countries by members and non-members.
Kingdom has the widest funding base and its accounts are at present quite healthy. SCA NZ has a good funding base but from only a small proportion of the Kingdom's population.
All stand to benefit from the continuation and improvement of a good Internet infrastructure.
Taking all this into account, I propose the following funding ratio, expressed on a dollar-for-dollar basis, i.e. no exchange rate conversions:
For each $1 contributed by SCA NZ, SCAA contributes $4 and Kingdom $5.
Looking at it another way: for a given budget dollar requirement, 50% is met by Kingdom, 40% by SCAA and 10% by SCA NZ.
For the first year, including the one-off cost of a new sca.org.au server, I propose a combined budget of $2500. Unless hosting issues arise, it is anticipated that subsequent years would be about half that.
This could, in principle be done by an existing officer, e.g. Kingdom Webminister, Exchequer or Kingdom Seneschal, but they'd probably find it easier to appoint a deputy.
Key responsibilities:
The options below are presented in descending order of desirability as I see it.
What's involved:
Accepting Raoul's offer to administer and host the server at his employer's (Aerocare's) Sydney site, and providing a budget for him to buy a replacement server.
Advantages:
Represents a continuation of a volunteer-provided very low cost, reliable service, on hardware owned by Lochac.
Disadvantages:
There is a small possibility that the free bandwidth on offer at Aerocare will at times be exceeded by demands on the sca.org.au server. If it happens, this can probably be managed. But, if not, one of the remaining options will need to be selected, as the costs of buying additional commercial bandwidth at that location are prohibitive.
What's involved:
Hosting the new server with Raoul, but shifting one or two high-bandwidth activities to Williams's sca.org.nz server. Conservatively, this could be done by shifting just a few mailing lists.
A more extreme approach would be to ask Willam to take on all "sca.org.au" mail services, leaving only the web services with Lord Raoul. This would definitely solve the bandwidth issue and would also allow lochac@sca.org.au-style domain names to be retained, i.e. users wouldn't even notice.
Advantages:
Makes it possible to keep the sca.org.au server at Lord Raoul's site with no significant bandwidth issues. Means that, even if bandwidth issues arise, we have an option to keep the replacement machine in Australia, instead of shifting to a US-based host.
This approach is also available as a stop-gap while considering or implementing a move to a US-based host. (NB: this approach, and load-balancing in general, would be much easier to manage if we already had a non-country specific domain, such as lochac.org or lochac.sca.org in place).
Disadvantages:
Administrative work involved in the shift.
Higher administrative and server load for William. Possibly very high if the "extreme" approach were taken.
The conservative option would result in domain name changes for each affected mailing list, (e.g. shambles@lochac.org instead of lochac@sca.org.au), meaning mailing list users would need to learn new list addresses.
The extreme approach breaches the important principle of trying to spread administrative and hosting load as much as possible between two or more volunteer providers. Too much will be loaded on William and his server.
What's involved:
Outsourcing some/all email services to google.com, which offers a free or low-cost service which may match some or all of our mailing-list needs.
Advantages:
Makes it possible to keep the sca.org.au server at Lord Raoul's site with no significant bandwidth issues. Means that, even if bandwidth issues arise, we have an option to keep the replacement machine in Australia, instead of shifting to a US-based host.
Does not create higher administrative and server load for William, as his server would not be involved.
No domain name changes for mailing list/s and/or email addresses.
Disadvantages:
One-off administration work involved in the shift -- more than for shifting to William's.
Some user impact (downtime during the change, change of mailing list administration interface, probably new passwords).
Some flexibility and synergies of closely managing our own mailing lists and software is likely to be lost, e.g. lochac-announce to lochac@ auto-forwarding, the idea of mirroring/fallover of mailing lists from sca.org.au to sca.org.nz, possibly also the easily overlaying and managing the proposed lochac.org domain.
Dependent on third party software and service including any restrictions, quality issues, costs, advertising, user experience issues etc. which they may decide to build into their offering at any time.
What's involved: Source a second-hand server in the US with appropriate specifications (about $700), install it at HREF Inc, which is Webcentre's (my company's) colocation service provider (about $300) and pay HREF about $75 a month for the bandwidth Lochac needs. Annual hosting costs: roughly $1000/year.
Advantages:
Good bandwidth on a dedicated machine at a very reasonable cost in an excellent, well-serviced environment.
Disadvantages:
Involves paying for hosting/bandwidth which we should try to avoid IF we can.
Remote management always requires more administrator time. It can impose unexpected costs in the event of hardware failure. The initial hardware purchase and installation is also harder to manage.
Website access for our populace is slightly slower due to the US location, but this will hardly be noticeable.
There are some conditions attached to the ISP's offer which means it might require renegotiation if, for example, their server rack space fills up. While not especially likely, it is a factor. Effectively, they are providing a non-commercial rate for this and we will be simply contributing to their overheads.
What's involved:
Select a service such as "level three" on offer at http://www.tektonic.net/unmanaged.html (about $80 a month or $1000/year). This provides a huge amount of bandwidth on a virtual machine which should be capable of running everything sca.org.au needs. If more capacity is needed, more money will deliver it, relatively easily.
Alternate service possibility: http://www.layeredtech.com/virtual-infrastructure/virtual-machine/the-grid-layer-virtual-machine/.
Advantages:
From our point of view, limitless bandwidth.
No direct hardware cost, just a monthly fee.
Hardware failures are not an issue and easy "upgrades" are feasible.
Disadvantages:
Involves paying for hosting/bandwidth which we should try to avoid IF we can.
Remote setup and management issues etc. as for (4) above.
Website access for our populace is slightly slower due to the US location, but this will hardly be noticeable.
A virtual server is usually not as good as a dedicated server. When compared to having a physical server that we know will be up to the task, we may end up suffering service failures or having to pay more money for a more capable virtual machine.
Fees escalate rapidly for higher-spec virtual machines.
Quality of support, while good, is unlikely to be as good as the first four options.
If you have any questions or feedback about this document, please contact Bartholomew.