> I'm taking this as a two-edged sword....if Cobalt doesn't take
> action on our requests, we can make one kind of judgement. On
> the other hand, if we can't produce a list of the above, then
> we are creating a lot of hot air.
This thread has not received enough attention. After so much bitching
(and even wanting to sue Cobalt), I think we all should take the chance
offered and give suggestions. (I'm only getting to this now since I fell
about 700 messages behind...) I'm also going to try to give a good
example and spend some real time on this...
[Note after finishing this: I've included here 30 concrete suggestions,
all of which (duh) I think are good ideas. I hereby challenge Cobalt to
implement some of these and thereby help itself while helping us. But
never shall it be said that I don't put my money where my mouth is.]
-----
<rattle, rattle, cough, cough, start brain wheezing along>
1. The Qube-2 Firewall Administration GUI is BROKEN. ipfwadm using the
GUI does not work in cases where masquerading is being used. It took me
two weeks to figure out how to do it by hand in the CLI... and I didn't
take notes, and I had to restore from CD and now it's broken again,
damnit. :(
2. Include the Restore CD with the product in all cases.
3. Make @domain.com the default email address in all cases. I can assure
you that the people who *want* to receive their mail @www.domain.com are
in the minority.
4. Make the GUI run faster. This could easily be done by (a) reducing
the graphical complexity of it, or (b) reusing the graphical elements so
they don't have to be downloaded each time.
5. Make the *use* of the GUI run faster. Have 10 users each set up 50
domains in a row using the GUI, and take their suggestions on how to
improve the interface. This particular test should take about four hours
per user and provide a wealth of learning. See my answer to Steve Buza's
post for an example of how to make the DNS setup TONS easier. Do similar
things to the creation of users (do it by batches of 10, where blank
cells are ignored when hit OK, or something).
6. Make new versions of the GUI read configuration files *each* time
it's called. This will allow those of us who wish to modify files by
hand to do so without having changes overwritten by the GUI... indeed,
the GUI should take advantage of this. Removing this would allow more
experienced admins to stay with the Cobalt line rather than moving to
vanilla Linux boxes.
7. Secure the GUI with SSL, even with a self-signed cert. No server
administration should be done by sending data in the clear, ever.
8. Telnet is evil; SSH (free version, of course) should be installed on
every box. See #7: no server administration should be done by sending
data in the clear, ever.
9. The concept of data backup should change. The ideal world would only
require me to:
* Restore from CD.
* Install applicable patches.
* Select the restore function from my full backup.
* Find all my users re-created with correct data
including passwords.
10. Provide ongoing support and some modifications to existing software,
even if it's a damned appliance. The Qube-2 should not be using
bind-4.9.7 any more! Upgrade to 8.2.3 and change the DNS Parameters
page... not that hard. Similar concepts apply to other software
installed. Also, why the hell is there still a 2.0 kernel with ipfwadm
on there? 2.2.16 with ipchains, anyone?
11. Logcheck, PortSentry, and ipchains (see #10) should be installed and
functioning properly on all systems. Pay Psionic $1.00 per system for
God's sake, and help them improve too.
12. Shadow passwords are not used in the Qube2. Fix.
13. There should be a "Home Qube" sold for home users who connect to the
Internet via dial-up, cablemodem, or DSL. Big market. Many of us can't
shell out $1,000 for the *low-end* box.
14. For marketing: can't you use something more useful than 20GB drives?
Sell an option for 80GB for an extra $150, for God's sake; you'll sell
more servers. Take a cue from Dell. They'll upgrade anything if you're
willing to pay for it.
15. Install something that tries to crack my own passwords (if I
activate it, since it's a CPU hog) so that stupid me can get some help
protecting me from my own users who have "mom" as a password.
16. Do a reality check on prices. $5K for an XTR just seems way too
high, priced out of its market. I thought this was the low-end division.
Do you have any idea what I can *build* for $5K? Including classes for
me, and Plesk or something else...?
17. Issue patches far more quickly. Do more QA... but do it quickly. The
BIND fix came fast, but the ProFTPd fix was *not* the same story.
18. Issue patches that don't break machines. It's sad that people are
sometimes more scared of the patch than they are of the problem.
19. Support this user community better. Cobalt has no *real* official
presence on this list (yeah, yeah, unofficial whatever... pipe up), and
it would do wonders for your marketing if there were. You'd also collect
far more feedback and keep your fingers on the pulse of your market far
better.
20. Support this user community better. 11 days to deliver a message to
this mailing list? Please... my 386 can do better than that.
21. Support this user community better. Has it occurred to you to offer
$100 discount coupons to the person who offers the most helpful post
each month? Or somehow provide an incentive for those who offer help?
22. Support this user community better. Create a better way to archive
this and other lists, so that the accumulated knowledge is more readily
available.
23. Did I mention that there are several concrete ways in which you
could (and should) support this user community better?
24. Let me repeat someone's words from recently: We are NOT the enemy.
We are your PAYCHECK. And we have invested heavily in you, so we'd like
to see you succeed and thereby help *us* succeed. Listening to us is a
Good Thing (tm).
25. Take a lesson from Palm. Sell hardware all you want, but also sell a
bunch of CD's that will install RedHat Linux, plus the Cobalt GUI, on
whatever normal hardware. Sell the sofware alone for $500 like Plesk
does, and you will make a shitload of money on things that don't require
return shipping, don't have moving parts, can't physically break (or can
be replaced for the price of a $10 FedEx Envelope), can be shipped
overnight anywhere in the world or downloaded... see my point?
26. Personally, if you did #35 I'd buy that in a heartbeat, and I'd
never buy hardware from you ever again... which ought to tell you
something too.
27. It'll be a cold day in hell before I pay you $200/hour for technical
support. I'll buy someone else's machine first. Now, if you offered to
install PHP, MySQL, and PHPMyAdmin for $50, you'd make a fortune. Must
be at least 5,000 who buy this, for total income of $250,000... you have
*got* to be able to program this at a profit.
28. The main website on a Qube should be easily modifiable. I want my
own logo on that stuff, not yours.
29. If I just replace index.html with my stuff on my Qube2, the menu bar
goes to hell. Must be a better way.
30. I'd like to easily modify the user's default webpage. For example,
note that my users speak SPANISH! I'd be happy to translate... just help
me easily get to it.
I think that's enough for tonight. Feedback and comments from anyone are
welcome.
--
Rodolfo J. Paiz
rpaiz@xxxxxxxxxxxxxx <mailto:rpaiz@xxxxxxxxxxxxxx>