Hi Robert,
Robert Neuschul wrote:
[...]
One is -
in effect - asking the customer to bet the future of their
business/organisation on the integrity and robustness of the tool one
is recommending [Let's not get into discussion about this being true
for many other common products - we may know this but the important
point is that it's what such customers perceive to be true].
True. And with Openfiler let's take a look at what we're asking them to
bet this on:
1) ext3
2) Samba
3) Apache
4) Linux NFS
5) LVM2
6) mdadm
7) Enterprise Linux 2.6 kernel
Pretty much common fare. They've been betting the integrity of their
data on this list of technologies for years now. What we've done is
brought them together into a small, easy to manage solution that
provides the same functionality which the uninitiated currently pay
proprietary vendors a significant amount of money to get.
Whilst products such as e-NAS, EMC/Clarion, and the like aren't cheap
they are mostly just plug and go, with a relatively well known 'name'
and a fairly high level of street cred to back them up, and with a
fairly global coverage of support and service etc. Recommending and
supporting such products falls into the old "No-one got fired for
buying IBM" zone.
Sure thing. And no amount of effort we could put into promoting
Openfiler would ever overcome this. But then again anyone with enough
coins to afford EMC, NetApp et al would not be looking at Openfiler
anyway. Except perhaps in the capacity of a gateway to an existing SAN.
You now have the prospect of using Openfiler within a VMware virtual
machine - so you could give multiple departments their very own NAS to
play with at the flick of a (virtual) switch.
OF doesn't [yet] have that presence and reputation in the wider world -
which makes the risks perceived by potential customers even higher. No
matter what levels of support we as vendors and the OF dev team might
provide, it's that customer inertia that has to be overcome.
On the one hand a paid for support scheme /may/ be a reassurance to
customers; on the other hand it can also seem highly premature.
Premature how? I'm not being defensive I just need to find out, in
concrete terms, where exactly we're not meeting expectations so that we
can close the gaps.
I'm very sympathetic to Rafiu and the team's objectives for the product
but I'm highly sceptical about the commercialisation of a product [or
more properly the services surrounding the product] so early in the
development cycle.
Actually, Openfiler is 4yrs old (at least) now. And AFAICT from similar
NAS appliance software available today, performs pretty much all the
functions required of a solution of its kind. Some things we do even
better than the shrink-wrapped (and closed) commercial stuff you get
currently. In other areas we lag in features/functionality - but then
thats what you have release versions for :). In your view how long a
development cycle would justify its commercialization.....?
The short answer is that the annual [or even one-time] value to any
customer of any such support offering is in direct proportion to the
value of the data which will be stored and accessed on OF and the
business or organsiation's dependency on that data. Which is why
companies such as OnTrack and Vogon can charge very large fees for
recovering such data.
Many here are techies and could probably cope with problems, with some
prompting and help from others in the group. However for OF to be truly
commercial it would need to be out there in userland alongside MySQL
and Apache et al: that's a place where the user may not be especially
techie but simply wants the features and facilities OF can offer. Such
users won't jump to use OF in the present circumstance because the
product is 'unproven'
You revolve around the point of it being 'unproven' and I'm curious as
how you came to such a conclusion. The software is being used in mission
critical environments currently. What else does it take to get it
proven? You forget (or perhaps are unaware) that OF is built on top of
CentOS - which in turn is a rebuild of a Prominent North American Linux
distribution (you know who). We chose to use this as a base for a very
good reason, one which makes your point about OF being 'unproven' moot.
CentOS, and the distribution upon which it is based, arguably have more
of an installed base than any top-tier storage vendor you could name.
More people using it means more bugs being found and fixed, which leads
to a more robust solution.
Several times we've been asked about supporting filesystems or features
that are not provided by the base kernel in the distribution. While the
desire would tend to be to support everyone's feature requests and to
have the latest/greatest functionality, we temper our development in
order to ensure the very stability that you point out is required of a
solution that is as important as a repository for critical data.
and the fixes/solution & docs etc are still "too
techie"
Sure thing. The value-add is in the support and services offerings.
Detailed documentation would fall into that category. Folks who are
technical enough and know what they're doing really don't need anything
aside from minimal documentation in order to use OF. Having said that, a
full administrator manual was made available for Openfiler 1.1 and many
(not all) of the complaints and questions on the list were related to
items that could easily have been garnered from the manual or from the
release notes if people actually cared to take a look. The term RTFM
exists for a reason :).
[note that this isn't especially a crit, it's mostly the nature
of the product itself]. In short, for end users' it's a risk/reward
calculation that [I believe] doesn't currently support a subscription
model. That balance can only be changed by making OF easier to use with
other OSs [Windows etc]
What changes, specifically, would you like to see in Openfiler that
would make it easier to use with Windows?
and much more robust
There you go again ;).
and much better documented
so that end users can support themselves more easily.
If Rafiu is suggesting a subscription model for us techies and/or
resellers only
Not at all. I'm proposing a subscription model for users of the software
in commercial environments whose needs are being served by Openfiler and
whom have derived significant cost savings in the implementation and
management of their IT infrastructure as a direct result of Openfiler
being available to them. Much is being argued in the corner of Joe
Regular User but it seems to have escaped the conscious that JRU
probably doesn't have an LDAP server, doesn't know what a snapshot is,
and thinks iSCSI is the name of a fizzy drink.
[rather than for end user customers] then that's never
going to generate sufficient income to cover costs; the target has to
be direct revenues from end users - and that in turn means generating
perceived values and advantages which don't yet exist.
Thanks for your detailed comments Robert. I can't say I agree with many
of the points you laid out, however it's good to get your perspective on
the topic.
KR,
Rafiu Fakunle
Openfiler Project
_______________________________________________
Openfiler-users mailing list
[email protected]
https://lists.openfiler.com/mailman/listinfo/openfiler-users
