cancel
Showing results for 
Search instead for 
Did you mean: 

The current plan

Ianwild
Grafter
Posts: 3,835
Registered: 05-04-2007

The current plan

Hi Guys,
Another month has passed since I last went to write this update, and although things have changed since then, I think now we're locked onto a plan for hosting that will solve our current issues with the CGI platform and give us a good product to build on. I will happily give you all the gory details of the journey to getting here at some point, but I think we can save that for another post. For now, I'll summarise by saying that it's been about getting to a plan for our next generation hosting platform that works and has buy-in from everyone. I think we're there now, except for one important group - you lot. So, I want to give you now a brief summary of the discussion and ideas so far, let you know where we are and where we're not, and hopefully (now this has been given some real priority across the business) keep you updated about where we are getting to as we progress this project from a closed trial, through to an open one and finally to a solution.
Options
It must be a year now since we all decided that the CGI platform in anything like it's current guise was simply not going to cut it for Plusnet's long term hosting solution, and following that it's been hard to justify spending resources to improve the existing platform. We've spent that time looking for a replacement and have explored in detail options around outsourcing hosting or partnering with other hosting providers, rebuilding the platform from scratch, or doing something in-between. Whilst we are still in discussions with a couple of potential partners now, we've pretty much reached a consensus that we want to try and keep this in house, but use the most common architecture possible and not go too bespoke else we end up in the same place as the current CGI Platform.
So, the plan is to use a third party hosting solution but run it ourselves. This should give us more control over our product, but let us offer something that works well and is easily kept up to date. To achieve this, we plan to use the same Plesk platform we are now using successfully to run PAYH, a dedicated hosting supplier we acquired as part of the Metronet / Parbin purchase. Although this site didn't originally use Plesk, the Plesk solution has proved to be effective at running the service since we took over the operation.
The Plan
As you may have gathered, the plan is to replace the current CGI platform and all the services it incorporates with a service delivered solely through an upgraded version of our existing Plesk platform. This won't include our homepages hosting platform, which will remain the default webspace included with all our accounts. We plan to keep that running separately for the foreseeable future, possibly developing this into a web based secure storage solution (The netapp platform it runs on is very stable and highly resiliant). Apart from that then, the change will apply to all these things: 
- CGI Webservers and Shell servers replaced with Plesk based service
- MySQL servers replaced with Plesk based MySQL hosting
- Frontpage 2002 Extensions support withdrawn (It's no longer being supported by Microsoft and is no longer provided in Plesk)
- Standalone PHPMyadmin replaced by the version included with Plesk.
- Plesk site Webstats provided via Plesk site URL rather than portal
- Anything I missed! (?)
The idea in the first stage on the new Plesk platform will be to provide equivalent products to todays Hosting offering, ie 250Mb webspace, 7.5GB a month data transfer and double for business, 1 database etc etc. This will be the entry level offering, and during the trial phase this will be completely free of charge to all trialists. I say that because we may choose to charge new customers even for this entry-level product in the future (at a nominal rate eg £1 - £2 a month). Zen moved down that route, and the idea is that it provide better impetus for platform development and to cover the costs of running a quality service as a separate entity. These discussions are being had now so feedback is welcome. We've also thought about charging current customers too (Maybe just ones who don't already use CGI), but this would need a Product change and is something we'd only consider if it was something that provided value and which we could explain properly. I'd be very happy to hear input on this element of the migration plan especially as with some financial impetus I think it will be easier for us to do the migration (We can justify more support for migration and only people who cared enough to pay would want their site moved)...
Once people are on the new platform, any customer will be able to simply upgrade to a better product if it's available. For example customers will eb able to buy more webspace and bandwidth and we will supply the same reseller type products offered with PAYH. There will also be all of the Plesk add-ons we can offer easily such as Tomcat, and a range of application packs and hopefully things like SSL certificates and other features can be added later. Billing will be done against a customers broadband subscriptions and it will be done as an invoice line item so will be available to pay via Direct Debit etc.
Products and Commercial Stuff
It's worth saying at this point that the integrations with our existing systems will be:
- Account creation and Deactivation and possibly loginshare with the Plusnet portal
- Billing via Broadband account
- Domain Records - We will continue to use Plusnet Domain hosting, so mail etc won't change. Domains (either internal like www.username.plus.com, ccgi etc or hosted domains / sub-domains) will all be able to pointed at the Plesk platform, which will have the domains created in them automatically when this is done via new functionality in the current portal tool.
The rest of the interaction will all be done via the Plesk control panel (which will have a look and feel overhaul from it's current self). The idea there is that new features can be added to Plesk easily without us having to do more work and these will then be available to customers immediately.
We're still deciding about support, but for me I'd like to use the PAYH support team to offer support for the hosting element. This may well relate to the decision about charging, for example the £1 a month might well pay for dedicated hosting support, whereas without that we won't provide any support at all (Migration or otherwise).
The real area of contention for us about the plan is how we handle existing customers using CGI and a migration to a new platform like this. It's bound to cause loads of problems, and I really hope in this group we can work to find ways to mitigate as many problems in advance as possible. At the end of the process, I'm not even sure if we should move all sites, or if we should make that opt-in only etc. Again, thoughts are really appreciated at this stage as I'm building up the requirements documentation.
I've actually got loads more to tell you about, in terms of the work we plan to do here. There's platform resilience and Internal support improvements, testing platforms, lots of comms etc etc. I think though there is enough there to spark some debate and to make sure we're not missing anything vital on this one. Certainly everyone here is asked to play a part, and you have a very important role in all of this, as without a true understanding of customers here we could potentially really screw things up. I'm under no illusions that it will be easy, but I do think the end result could be well worth it.
Timescales
In terms of timescales, I'm thinking something along these lines is possible, however none of this is properly scheduled so it could change dramatically (As could any aspect of these plans based on many things).
- Closed Trial commences (Invitees to this forum - please feel free to suggest more people to help) - Till Mid September
-- Manual account creation and no integration - This week
-- General testing of platform and products so we can capture further requirements
-- Moving of existing sites on CGI platform to test them
- Open Trial of current platform during September
-- Proposed products will be available and the solution will be a viable hosting alternative to CGI for people prepared to manually create a PAYH account and move their site.
-- Testing of migration generaslly, site moves and how we do DNS properly will run into October.
- Test site migrations with opt-in users will be done during October
-- We'll be moving people to the platform and helping with migrations
-- We'll be doing skinning and improving the tools on the platform by then.
-- Towards the end of the month, we'll be testing billing via broadband subscription
- New customers being added to this platform instead of CGI by November / Early December
-- Automated account creation etc.
- Final migrations completed and oId platform shut-down by end of the year
Again, those timescales are plucked from the air so don't quote me on them just yet - They are merely what I intend to propose internally once the plan is agreed, to give you an idea of my thinking. The most important things for me are that we do it right and don't make compromises that will cripple us later, as happened with CGI, so it needs to take longer it will. However, the advantage of a plan like this is that we should be able to let customers who want a stable and better hosting platform than we provide today get that pretty quickly, even if there are some manual processes involved initially in setting them up.
Sorry for the long post - I will look forward to your feedback.
Cheers,
Ian
20 REPLIES
avalon
Grafter
Posts: 361
Registered: 05-04-2007

Re: The current plan

Ian
Any figures on the number of sites involved in total and also those currently actively using the current ccgi?  i.e. sites with actual web visitors and excluding personal storage/software test space, abandoned projects, etc.  I'd guess that the active sites will be less than those with files in ccgi space and that the most active users will be more capable of transitioning to a new hosting platform.
Some will be straightforward to move and some will be much harder, perhaps having been set up by someone else, or where the knowledge has been forgotten on how it was done.  Unless a lot of PN resource is devoted to assisting the move the December turn-off of the old servers sounds rather optimistic given it's now mid August.
I'd say you should be looking at providing what is currently available FOC, but sell addon space/bandwidth/databases/backup services at a modest profit.  It'd be hard to convince many people to switch if the ccgi service they're getting for free just now is doing what they want, and ultimately you want everyone to be keen to put in the effort to transfer over so you can turn off the old one rather than fighting you tooth and nail over it.
Community Veteran
Posts: 1,571
Thanks: 3
Registered: 13-04-2007

Re: The current plan

One of the problems in the past has been when the servers went down and sites were down for 2 days and some were lost. So the important thing from a customer point of view is SLA's and backup.
In the past it was tough luck if the servers went down and there was no backup by Plusnet what is needed is to use the website commercially so you dont expect it to go down for days or be lost without backup
Ianwild
Grafter
Posts: 3,835
Registered: 05-04-2007

Re: The current plan

Hi Guys,
Although there are around 10,000 registered accounts on the CGI platform, the vast majority of these are effectively unused. It is hard to say exactly, but based on all the information we have we estimate that there are somewhere in the region of 300 - 400 customers actively hosting websites on the CGI platform currently.
One of the key points about the PAYH brand is that it's already running with active support and back-ups, and downtime is pretty minimal. Plesk has some inherent resilience challenges in it's design, but even if we completely lost a machine, group of machines or even one of our data centres, we'd be able to have all the sites hosted on those servers back up and running very quickly indeed. Uptime on the current platform is exceptionally high.
Ian
Superuser
Superuser
Posts: 2,485
Thanks: 192
Fixes: 5
Registered: 06-04-2007

Re: The current plan

Only managed to have a quick read at the moment, but have 2 points to mention.
1st one is what happens when people leave Plusnet.  Could the hosting account be changed to a normal PAYH account, and exist without the customer being a broadband customer?  ie, say I've got a live website hosted here with my broadband connection (either free or your small charge), maybe with extra disk space and/or bandwidth.  I decide to move my broadband to another supplier, could I have the option of keeping the hosting for the fee that would be charged if I'd done it in PAYH in the 1st instance, with out me needing to do anything with my files, settings  or anything else and without any downtime.  What would happen to email if you did this, since that isn't set up in plesk?
My second point is, if you didn't migrate everyone from the cgi machines and did keep them running, what would happen if a security hole was found in PHP4, which would be unsupported, and hense not fixed?  Presumably the only option would be to upgrade PHP (which is one of the things you're trying to avoid by replacing it), or shutdown the service with little notice.  Also I'm sure it would be in your intrest not to have the cgi platform hanging around with a few users on it.  Surely it would be better to scrap/redeploy the hardware anyway.
I'll try and revisit this a little later incase I find anything else to bring up.
Oh, and thanks a lot Ian for your post.
Phil
Ianwild
Grafter
Posts: 3,835
Registered: 05-04-2007

Re: The current plan

Thanks Phil,
It's certainly no problem if people want to keep their hosting with PAYH after leaving. There would be two options:
- We already have hosting only accounts that allow us to bill for everything but broadband and mean you could keep using our Email with Broadband from another broadband supplier. I think prices would need to start from about £3 a month for that.
- If people move their domains over to PAYH in full at the point of cancelling Broadband, there is no reason you won't be able to have one of the existing PAYH accounts and the Plesk based email set-up without any issues at all.
We definitely would need to address a PHP4 issue if one arose. We can't physically upgrade the boxes, so the only real answer would be to accelerate the plans. That's why there is priority in getting the plans in place, although hopefully we'll be able to offer a reasonable period for migration so that no one gets stung by the change.
Regards,
Ian


Superuser
Superuser
Posts: 2,485
Thanks: 192
Fixes: 5
Registered: 06-04-2007

Re: The current plan

My second point about PHP4 security was more in relation to what you said here:
Quote
I'm not even sure if we should move all sites, or if we should make that opt-in only etc. Again, thoughts are really appreciated at this stage as I'm building up the requirements documentation.
If you made it opt-in only, that would presumably mean there are sites that would be left on the old platform, or where you meaning that those that don't opt in will have their sites killed when that old platform is taken out of service?
Phil
Ianwild
Grafter
Posts: 3,835
Registered: 05-04-2007

Re: The current plan

Yep - I was meaning the sites would die when we switched off the old platform.
I must stress that we'd control the way we did that and provide notice etc. We'd also have the data available for restoration on the new platform by the customer (or potentially by us or a partner for a fee).
One of the success criteria's for this project is that it can't impact customer satisfaction or the helpdesk, so no decisions in this regard will be taken lightly.
Kind Regards,
Ian
avalon
Grafter
Posts: 361
Registered: 05-04-2007

Re: The current plan

Quote from: Ian
Although there are around 10,000 registered accounts on the CGI platform, the vast majority of these are effectively unused. It is hard to say exactly, but based on all the information we have we estimate that there are somewhere in the region of 300 - 400 customers actively hosting websites on the CGI platform currently.

Blimey, a bit of a difference there then.  Perhaps concentrating on the 400 and offering detailed advice for self help migrations, or a personalised migration assistance service for a small fee would be a good compromise?
I doubt there will be much scope to sell additional services to the other 9,600 so a more automated approach to just storing/transferring their data could be offered for all those ccgi accounts that aren't in active use.
When you see the amount of unused accounts I can understand why you'd be interested in charging a nominal amount for it, even if just to discourage people from keeping it if not using it.
Ianwild
Grafter
Posts: 3,835
Registered: 05-04-2007

Re: The current plan

Yep - That's precisely our thinking. It is actually hard to tell what active use is - The 400 number comes from stats that look like someone other than crawlers is indexing a site - ie data transfer of over a couple of meg in a month and a few other factors like SSH and FTP use. Lots of people though might have content they want to keep and we need to strike the right balance. That's my task for next week...
By the way chaps, we're looking at the 20th now for the PHP5 upgrade on the new platform. The guy who was doing this has to do some other rollouts at the same time (in an overnight slot) and it makes most sense for us to wait till then.
Ian
Community Veteran
Posts: 3,789
Registered: 08-06-2007

Re: The current plan

Ian,
In my eyes one of the reasons that the ccgi server is underused is the level of expertise required to actually install anything on it.
I note that Plesk provide a Sitebuilder package, which includes many popular packages available to the EU at the click of a link.  Packages such as a forum, gallery, guestbook are all desired.
Does the PAYH product line use the Sitebuilder package, or is it one of the other Plesk product offerings?
IMO, offering Sitebuilder is a good way to promote the PAYH product to more casual users and expand the usage of the site.  Users would be more inclined to pay an additional charge for a system where it is mostly a point-and-click process to set up whatever they like.
Of course, the more demanding user will still have the access to generate whichever type of site they like.
B.
Ianwild
Grafter
Posts: 3,835
Registered: 05-04-2007

Re: The current plan

Quote from: Barry
In my eyes one of the reasons that the ccgi server is underused is the level of expertise required to actually install anything on it.
I note that Plesk provide a Sitebuilder package, which includes many popular packages available to the EU at the click of a link.  Packages such as a forum, gallery, guestbook are all desired.
Does the PAYH product line use the Sitebuilder package, or is it one of the other Plesk product offerings?

It doesn't at the moment,, but I'm sure we could offer these add-ons for a small extra cost (Sitebuilder costs about 11p a month per site, which makes it impossible to give away free on most of our products). I'm spending some time today playing round with what I think these costs might be and detailing more of the plan - I'd welcome your feedback.
Quote
IMO, offering Sitebuilder is a good way to promote the PAYH product to more casual users and expand the usage of the site.  Users would be more inclined to pay an additional charge for a system where it is mostly a point-and-click process to set up whatever they like.

Absolutely right, I think that has to be the way to do it, so we end up with something like this:
Residential Options:
1. £0 per month - Community Support Only*, 250Mb Storage, 1.5Gb Data transfer per month, 1 MySQL DB.
2. £2.50 a month - Telephone / Ticket Support (from the PAYH support team), 250Mb Storage, 1.5Gb Data transfer per month, 3 MySQL DB, Application Packs + Site builder.
Business Options:
1. £0 per month - Community Support Only, 500Mb Storage, 3Gb Data transfer per month, 1 MySQL DB,
2. £5 per month - Telephone support, 500Mb Storage, 3Gb Data transfer per month, 3 MySQL DBs, Application Packs + Site builder.
We can then (later for some of these) have competitive and flexible chargeable options around TomCat, Fixed IP Addresses, Additional Bandwidth, Additional Disk Space, Additional databases and SSL Certificates.
* By community support, I mean that the support team won't offer help with using the web hosting platform, but they would still be available if there was a platform outage. That is the same as we have today with the current CGI platform.
I should stress that the above is just a rough summation of my thinking, but I would welcome any feedback you have.

Ian
Community Veteran
Posts: 3,789
Registered: 08-06-2007

Re: The current plan

I think it's important not to "take away" the value added services such as free webhosting so the entry-level price points are perfect from that point of view.
The upgraded accounts are certainly in the right ballpark for professional hosting packages, but I would like to see some firm level of commitment as to the level of support available on those options. 
Also, PHP isn't the only hosting application software available - there have been requests for Ruby for example which may bear some investigation.  What I'd like to see is not having to provision "special" servers just to accommodate things like Ruby but have it available out of the box.  Good to see Tomcat as a potential too, although anything offered really will have to be planned so that continuing support will be available throughout the product lifetime.  Regular upgrades to PHP/Tomcat/Ruby *need* to be factored into the equation so that we don't hit another obsoletion process in the future.
Addons such as SSL certificates and extra bandwidth/disk space are also welcome additions to the product line - something which has oft been mentioned.
There were reservations about the scalability of Plesk raised - have these been investigated thoroughly? 
That's all for now, but if I think of anything else be sure I'll return Wink
B.
Superuser
Superuser
Posts: 2,522
Thanks: 920
Fixes: 8
Registered: 10-04-2007

Re: The current plan

The package is beginning to take shape Ian..  The entry cost is OK as a carry over for current users, and the paid for charges look OK for small Home / Business use.
I'm with Barry on the need to support more than just PHP,  Ruby has been requested a number of times in the past.  I guess the loss of User interest in keeping that particular pot boiling is because many of the more savvy users have already migrated their web development to another Host.
I'd also like to see significant progress made on how the Hosted accounts are managed for transgressing either the storage or bandwidth limits.  The current method of removing the site with no notice is not well received by users, there must be a better way to control and protect the network and keep the end user sweet?  OK so something needs to be done, but at least keep the site on-line to allow the user to rectify the problem first before removing it?
Could I ask that you also review all of the current PUGIT items on web hosting, and see how many of them can be cleared in the new Hosting plan - not all necessary on day 1, but some sort of committed road map would help sell the transition.
I guess PUG and the Mods need to get ahead with writing some "How you do it" tutorials for some of the more obvious packages.
M
 
aaronbennett
Dabbler
Posts: 15
Registered: 04-10-2007

Re: The current plan


Ruby 1.8.5 is installed. Rails is installed.
I've implemented something very basic Ruby on Rails style via the command line; and also put up a .rb CGI as a test. Both worked fine.
My knowledge of using this language is limited but I'm happy to run further tests if somebody would like to point me in the right direction.