Date: Thu, 18 Mar 2010 12:19:26 +0000 From: Bruce Simpson <bms@incunabulum.net> To: freebsd-xen@freebsd.org Cc: James Smith <James.Smith@m247.com> Subject: Re: XenTools development for FreeBSD Message-ID: <4BA21A4E.2040108@incunabulum.net>
next in thread | raw e-mail | index | archive | help
Hello everyone on freebsd-xen!
I should like to share with you all some developments!
There is clear demand for this out there, we just need to negotiate how
FreeBSD developers get compensated reasonably for their time on this
project. To my knowledge, most of us in FreeBSD aren't independently
wealthy, so we do need to charge reasonably for our time, when the work
goes beyond that which a volunteer could reasonably be expected to do.
What's happened:
I visited M247 Ltd in mid February. They are a young-ish datacentre
operator who have moved their operation from London Docklands to
Manchester/Stretford, just off the Bridgewater Canal, to reduce their
overheads.
All James and his organisation wants is for FreeBSD 8.x releases to
work fine out of the box in Citrix XenServer, and for live migration to
work. It's that former part which carries the high price tag, because of
the QA needed on the existing code base, to get back to stability.
So I have simply tried my best to get things going, and sadly that
has not been enough in this instance. Normally I'm a contract R&D guy,
not a salesperson, so I apologise for my brash job in trying to convince
James of what actually needs to be done.
Whilst they were interested to talk about getting FreeBSD/xen
working in their infrastructure, I pitched this work purely on a
commercial basis, at a very reasonable hourly rate for the market I am
in, to give them some idea of what the real costs of retaining a kernel
developer on a very specific project amount to.
The outcome of this:
Based on the reaction to date, I believe this has given James
(Cc'd) the frighteners, learning what things actually cost in the
software business... !
But to be fair, software is not M247's main line of business, and I
certainly don't expect every company I visit to be able to evaluate an
expert's work, nor necessarily being able to fund it completely out of
their own pockets. Companies get nervous about employing lawyers for the
same reason; hourly rates for experts are high.
These discussions haven't taken place in the open, partly out of
respect for commercial confidence and ongoing negotiations, which have
now broken down. In my zeal to close, I quoted on the basis of what the
customer needed, not what they said they wanted, and I must fall on my
sword for this.
In my view, a casual hourly arrangement would have worked fine for
this (as this is how the majority of my US clients do business with me),
and we might even have been finished by 2 weeks from now, although that
is easier for me to say with a src/ checkout on my screen.
What's actually involved:
Porting Citrix XenTools to FreeBSD is relatively trivial if all
the other pieces are in place, it just isn't going to work for this
client, without the QA and fixes I was actually quoting them for. That,
is purely a matter of engineering. That, and the fact of the significant
differences between open source Xen, and the XenServer product.
Hence this message in the open. I have kept Kip, Adrian and Robert
in the loop about this. I have also been chatting with Justin Gibbs, who
is now actually doing much of the QA polishing on FreeBSD 8.0 for his
employer.
The FreeBSD Foundation CFP closed on 1 March, and so I was eager to
guide M247 towards this as a pot which could match their funding for
getting QA on the FreeBSD/xen port; Now, we have an NLNet funding
deadline up on 1 April. It is now the 18th of March, so time for that is
closing fast.
Unfortunately without their buy-in, there is little else I can do.
I don't manage projects or write project plans for free usually, and I
wouldn't be expecting full-blown 3GPP daily rates for, what is by
comparison, baby steps in terms of what deliverables and resources need
managed.
And my position:
Anyone who chooses to pick this up, I am happy to forward you my
code bookmarks (from reverse engineering XenTools), and details of the
test procedures I used to determine FreeBSD/xen wasn't ready to go for
this client. I have a copy of the Xen Hypervisor tools book, which I
could pass on via donations@.
Other prospects on my radar are not interested in Xen. For the
moment, I am focusing on other things, as I seem to be attracting strong
opportunities at the moment, but I wanted to summarise where we got to,
in case anyone else chooses to get involved with this.
cheers,
BMS
P.S.:
If we had a FreeBSD consultancy shop established in the UK, as a
private enterprise, it would probably have been a simple matter of
sending a salesperson to connect with James, and give him a firm figure
and timescale. However, we don't have one of those, and Kip has been
doing everything off his own back, as far as I know.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4BA21A4E.2040108>