[Wolves] Converting SBS2000 to a linux based server

[James Turner]

On Thursday 11 Nov 2004 14:02, Chris Ball wrote:
> mainly the work of an idle mind.
>
> Is it worth moving away from Active Directory and Server2000 towards a more
> linux approach to my Servers?

Maybe. From a technical point of view, Linux is easily capable of performing 
an equivalent role to a similar or better standard, depending on specific 
needs. Here are some examples of Open Source alternatives to various SBS 
facilities:

(use of fixed-width font recommended to view table)

Microsoft based solution                Open Source (Linux) based solution
------------------------------------------------------------------------------
File and print services                 Samba
Active Directory                        Samba (acting as NT Domain Controller)
Exchange                                Sendmail, Exim or Postfix
Exchange Outlook Web Access             SquirrelMail or Horde
SQL Server                              PostreSQL or MySQL
Internet Information Server             Apache
Active Server Pages                     PHP, Python, Perl, CGI and many others
Shared Fax Service                      Hylafax or Mgetty-Sendfax/Getfax(?)
Internet Security/Acceleration Server   Squid, SquidGuard, IPtables
SharePoint                              Gluecode or Metadot(?)
Routing and Remote Access               PPPD, Linux Kernel - IPtables, IPsec
FrontPage                               Mozilla Composer, Quanta, text editors
Internet Explorer                       Whatever web browser you like
Outlook                                 Whatever POP/IMAP/SMTP client you like

> I understand it will be more hard work to set up, but for a small company,
> would it be worth it?

The cost/benefit breakdown would depend on the specifics of the company, which 
we obviously don't know anything about. What are the current and future 
requirements? Which of the SBS components are in use? A detailed feasibility 
study may be needed to answer the question fully.

Some of the issues you may want to consider (in no particular order) are:

      Are there any office politics or prejudice-based reasons why one type of
      IT solution should be chosen over another?

      What's already in place? How much is likely to cost to change/upgrade it
      now or in the future? What are the perceived benefits? Is it worth
      making a change?

      What risks are involved, are they worth taking and will they grow/shrink
      in the future?

      Are routine sysadmin tasks easy for staff to perform, without too much
      risk of costly errors? If desired, are outsourced/managed solutions
      providers available to provide a system administration/support service?

      What is the proposed product's record on reliability? What resources
      (time, staff) are needed to cover routine maintenance and support
      tasks?

      In financial terms, how critical is reliability to the continued
      running of the business? What is the financial cost of an outage or
      malfunction? Are high availability features available, and at what cost?

      Do IT staff have the skills to be able to support the product? Would
      training be needed?

      What level of security does the product provide? What are the
      consequences of a security breach in terms of financial cost and
      business continuity?

      Is it easy to make/restore backups? How long would it take to restore
      a working system following a major hardware or software outage, and
      what costs are associated with this? Have you actually tried this in a
      test environment?

      How much does it cost to license an additional desktop machine?

      What desktop operating systems are supported?

      How much does it cost to add a desktop machine above any hardwired limit
      within the product? (i.e. how much does it cost for a 51st machine with
      SBS?)

      How much will it cost to upgrade to a future version of the product?
      Will new client licenses be needed for every desktop?

      Does the product have any built-in limits to scalability? Is there a
      smooth and cost effective upgrade path to higher-end products?

      Is the product based on genuinely open, widely supported standards that
      will facilitate broad interoperability with third party applications,
      etc?

      Can extra facilities be added to the base product fulfil future needs?
      Are they readily available and how much do they cost?

      Is the product (or a successor) likely supported for a long time into
      the future? Does this depend on the needs of the users or the vendor?

The problem with many current Microsoft-based products is that when the sales 
people demonstrate them they appear (to the untrained and uncynical eye) to 
be fantastically easy and simple to run whilst providing unique and amazing 
levels of functionality, performance and reliability. It's only when a 
solution has been purchased and the poor folk "on the ground" have wrestled 
with it for a few days/weeks/months that the grim reality starts to sink in.

I've heard too many false or misleading promises (like "high level, 
government-approved standard of security" on NT4, then 2000, then XP) to take 
much of the sales patter at face value any more.

Anyway, enough waffling and bitter MS complaints for one night (from me 
anyway). Let us know how you get on!

Best wishes,

James