Enterprise Linux: Are We There Yet?

Simon Crosby writes " Network Computing is running an special report on Linux in the enterprise. It evaluates strengths and weaknesses of Linux useage in the enterprise. It also discusses perceptions, roadblocks, security, clustering and other Linux enterprise issues."

Linux in the Enterprise...

...will definitely give Picard an advantage over the Borg.

Re:Linux in the Enterprise...

Linux is not ready for the Enterprise. There is not a single voice-controlled app for any of the mission-critical functions of the Enterprise. Conspicuously absent are warp core control, phaser bank activation, interstellar navigation, transporter operation and the all-important self destruct sequence. Until these and thousands of other important apps are written and deployed, Linux will just be a toy in the Enterprise.

Lack of Apps

Are we there yet? certainly not. Linux has shown in the last few years that it is an alternative, although it still cannot compete in every Aspect with commercial Un*x Systems. (Especially Performance)

The greatest drawback for using Linux in your Enterprise is not the Performance issue but lack of Applications. Many Porting efforts are still beta, (Or do you consider Oracle to be stable on Linux?) or simply not done.

It is still difficult to convince the big software firms to actually consider Linux as an alternative, especially in the Enterprise computing field.

There has still much lobbying to be done.

Jeff

Re:Lack of Apps

While it's true that not all the un*x-based apps have been ported, most of them are there. Examples:

• Oracle -- which is rock solid, and you can get support contracts from Oracle for the Linux platform (the company I work for uses Oracle/Linux in a mission critical environment and it works just fine).
• All the other big databases (IBM DB2, Clustra, Informix, Sybase, etc)
• Lots and lots of movie and animation production tools (Maya, Entropy, etc). Linux is really popular now amongst graphics and movie professionals.

The list could go on, but you get my point. What's more important are the advantages of Linux: superb development tools, open architecture, world-wide support, and so on.

On the performance front, it seems that you're not aware of the fact that Solaris (and other unices) scale up so well to high-end boxes at the expense of low-end performance. In case you didn't know it, the Linux kernel smokes away Solaris (in terms of syscall latency, throughput, response time, network performance and a couple other points) on servers with up to 4-8 CPUs.

In certain areas...

I would go so far as to say that Linux is now the choice solution for enterprise web servers - Apache is all its glory, etc.

Doubly so, given IIS's press lately.

Are we there? Will anyone ever be there?

Linux is as ready for the enterprise as any other offering (including those already considered to be enterprise platforms). Work with OS/400 or even a commerical Unix for awhile and you'll soon find out that most of Enterpriseness is political. (Some is also legal... if you run into a snafu with kernel 2.6.1, who can you sue??).

Having worked with a large number of platforms, I can confidently say that Linux is up there with the best of them. It's not a leader in all niches (such as home computing or workstation graphics) but it's more than enough for almost all server, developer workstation, and terminal areas. In fact, I will go as far as saying that I would fully trust kernel 2.4.16 + ext3fs for almost any task with in reason.

Linux is at the point where the limiting factor is end user software, not the OS or its libraries. It is time to deploy linux far and wide. Linux will never be the perfect OS... no OS will ever be. It is time to deploy linux.

"who do you sue?"

could this question please die? considering the realities behind shrinkwrap licenses and ever dedicated support contracts, you can't sue anyone. and even if you could - how can you hope to win? the us gov't sued microsoft and look at how victorious it was?

a better question is: if things go wrong with widget x, what are my options to get it fixed? with closed s/w, the only option is the vendor you got it from (and really, knowing that, do you want to sue them?). with free software you can use your vendor, another vendor, your own staff, or private contractors (and knowing that, you could feel free to sue your vendor; assuming they failed to live up to their support contract).
note: i'm assuming in an enterprise situation you'd have some sort of support contract with ibm, redhat, microsoft, suse, sun, linuxcare, apple, etc.

"could this question please die?"

No, because the question itself is wrong, and is really a red herring.

It isn't "Who do you sue?" because instead it's really "Who can I blame and send the heat somewhere besides me?" The IT management structure will take heat for any service problems, but with a Microsoft solution they have the perfect blame target. Between "Everybody uses Microsoft," which absolves blame for having chosen them, and the fact that Microsoft is essentially lawsuit-proof, between their EULA and size/tactics, things are nicely diffused. Doesn't keep the systems up an running, but at least you're suffering in the same boat with everyone else, and there's the general, "Nothing can be done any better," to protect you.

Contrast that with Linux and outsourced support. First off, you've chosen something different, and hence inherently risky. Second, your outsourced support is probably less lawsuit-proof, and therefore maybe something might actually have to be done, rather than sighing in resignation.

Also contrast with Linux and internal support. Now you're to your own resources, and directly and immediately responsible for anything that goes wrong.

Note that NONE of this says a single thing about service levels, outages, or whatever. It's merely about adequate 'diffusion or responsibility' to keep the IT peoples' jobs protected. Microsoft provides a great 'responsibility diffusion sink,' one of the best at that.

Re:Are we there? Will anyone ever be there?

(Some is also legal... if you run into a snafu with kernel 2.6.1, who can you sue??).

You sound like you've got a good view of the issue, but this sentence cries for rebuttal. When, oh WHEN, will pople stop parroting this nonsense? Any CIO that uses this as an argument against OpenSource/Free software is a moron. I challenge anyone, anywhere, to give evidence that anyone has ever collected a single penny from suing a mass-market software maker for shoddy code. If MS didn't lose their shirt over putrid crap like win3.x or win9x, with it's dll-hell and semi-annnual re-install schedule, how can anyone get sued?

It depends on the enterprise they're talking about

Is this TOS, TNG, or ENT? I think Linux could make a lot of headway in ENT or even TOS. It's got years before it can compete with the system in _The Next Generation_.

marketing

my thought is that linux will eventually pick up, but it's going to take a while. Why? Linux works great and people will eventually realize that, but there really isn't much marketing as far as I can tell. IBM with it's Peace Love and Linux campaign has probably done a lot, but compared to MS . . .
*shrug*
Once linux's PR is going well, poor MS will probably be in for the fight of it's life . . .

MS Word

I think MS Word interoperability is perhaps the single most important barrier limiting companies from changing to Linux. Other Office products such as MS Excel and MS Powerpoint are also important.

You should remember that it's not just necessary to have some semi-lousy import filters to Linux word-processors, but also have 100% compatible export filters. It's practically impossible to make a transition in any company that has to communicate with an existing MS Word user base. And that is the case for almost any companies and public administration.

And 99% doesn't do, it must be 100.000%. If there are even small incompatibilities, you have to use genuine MS Word -> MS Windows.

StarOffice 6.0 beta (same as OpenOffice build 638c) has some compatibility in basic formatting. The older StarOffice 5.2 has, in my experience, much better MS Word compatibility, but it also breaks up quite quickly. However, its Excel compatibility is worse than with SO6.0b/OO638c.

KOffice (1.1) is not even worth mentioning with regard to MS Office compatibility. Its Word import filter simply strips all formatting, and it doesn't have an export filter.

I work in an IT company, doing purely Linux work, but have to do all documentation, communication, and administrative tasks with MS Office. I was able to use StarOffice 5.2 for a while in some tasks, but can't rely on it completely. The situation really sucks.

Office Apps != "The Corporate Enterprise"

I think MS Word interoperability is perhaps the single most important barrier limiting companies from changing to Linux. Other Office products such as MS Excel and MS Powerpoint are also important.

Desktop office applications are a noticable but small part of "the Enterprise" and NOT the main point of the original article.

"Enterprise" usually refers to the core applications running in the corporate data center. Inventory, payroll, order processing. Applications where downtime costs $$/minute. Applications where "No application"=="No business".

Linux is making gains in these areas. The adoption rate appears slow because

1. It is slow. "Bet the company" decisions are always slow. Implementation is slow. Anyone remember how long it took Windows NT to break into corporate data centers? (Many would argue it still isn't ready)
2. Companies don't always consider what is a "mission critical" application. Areas where Linux excels - web, mail, dns, and many other RFC-based services, for example - may not be viewed as "critical". At least until the boss wants to know why the corporate web site is down (nmida) or the email system is hosed (badtrans). Then we get something like the Giga Group recommendation to use anything but IIS.
3. Companies see this as a competitive advantage and do not want to discuss it. The big NYC financial firms are a good example.
4. Consulting firms need more linux experience. Many enterprise customers rely on the IT consulting arms of the big system integrators and consulting firms. If these outfits push something other than Linux, something other than Linux gets proposed. Do they get incentives from MS, Sun, IBM, HP, CPQ...? Maybe. Anyone pushing Linux like that? Not yet.
5. CIOs don't always know what's running. I've come across repeated examples where the top managers swear "Linux isn't allowed" but there are stealth, pilot, and production deployments all over the shop. The file sharing and print system runs Samba on Linux and "just works" (and isn't considered "enterprise" until a key document is needed).

Penetration of Linux could still be better, of course. We need better support from enterprise management and backup systems. We need more "mind share". This article helps.

Desktops remain a problem. Out of sight, out of mind. Windows is in everyone's face every day.

We may not be enterprise, but...

It has been, and will continue to be right for us:

I am one of two people who run the MIS office at a small technical college. We support everything that plugs into the wall for ~500 staff, faculty, and students. We're not huge, but we keep busy :) We run a Windows 2000 Adv. Server domain that seems to run well for the most part but acts... haunted, as if it's just scared to run stable all the time.

Recently we decided to replace an old outdated MS-Proxy server (don't yell, I know) with a Squid server. Not because squid is more stable, but it was damn near impossible to monitor MS-Proxy cheaply and effectively. We farmed around for sollutions but they were all over $1,500 for anything decent. Any of you familiar with Squid can finish this story yourself. For the others, now we can monitor at any depth all we want for $free (and of course it's more stable).

Last night I sat that same server up with MRTG to monitor all of our switches and the network interfaces on all the servers in our farm. Oh, and it's also a syslog server, watching over the the network printers and the server farm itself ( we have Event Viewer in Win2k setup to dump to a CSV file and then push it to syslog). It's also a planned backup www/ftp server and what-ever-else-is-needed server.

Proxy, MRTG, better monitoring then Event Viewer, syslog, www/ftp: all on one box, at zero software cost. Since it's Linux I know I can trust it to keep up. And I'm not saying that because I'm wearing my shebang(#!) hat and my tux hoodie, I'm talking from experience. You can't beat it.

That's why Linux is here for us now, and it's because of those qualities that Linux will continue to grow with us in the future.

The definition of 'Enterprise'

Somebody please define it, in the context of software. What makes linux any less "Enterprise-ready" than NT, AIX, Solaris, etc?

Does it even have a real definition, or is it just nonsense like the term "supercomputer"?

Re:The definition of 'Enterprise'

"Enterprise" has two components, first is the relatively straightforward "Core Application" of a company, which might be some sort of ERP system like PeopleSoft, or some other commercial produce. For many companies, this will be an amalgamation of custom programs written over many years. These applications are typically based on some form of transaction processing system (e.g. CICS on the Mainframe, Tuxedo on Unix, or even database-driven transactions ala Oracle, DB2, etc.)

The second, and more critical part of 'Enterprise' is the nature of the computing service. Generally any outage is measured in dollars per minute or hour. It's not unusual for a large company to face severe monetary losses for even slight outages. Think millions of dollars an hour (or even per minute). This measure tends to be a little slippery, but with some analysis a pretty solid figure can usually be determined.

For some enterprises, Linux might make complete sense (e.g. Google). For others, the potential of saving a few thousand or hundred thousand in licensing costs pales in comparison to the probable re-training, new hardware, and "potential" instability of moving to Linux. If you've got something that works, why fix it?

Given the above, even if all of the big 'Enterprise' vendors port their software to Linux, you're not done. Linux clustering in a business context such as Solaris, AIX, and (in the good old days) VMS provide would be one stumbling block. The lack of high-end hardware is another -- and yes I know that Linux runs on anything from a PC to a SPARC server to a S/390 mainframe. In reality, you're unlikely to drop $2million on a big Sun box then load Linux -- you'll want to take advantage of Solaris's dynamic partitioning and other proprietary hooks.

Loading Linux on diverse old hardware makes business sense -- turn that old Sun box into something useful. It doesn't make nearly as much business sense when buying a new non-intel server, since the license fee of the OS (if any) is negligable compared to the overall value of the system in the 'Enterprise'.

Over time this is likely to change, since Linux represents a constantly improving and freely available system, vendors will start adopting it as 'their' OS. IBM is an early starter here, but the process will take time. And like a battle of attrition Linux has the advantage over time, since it is constantly improving (for free), while commercial vendors have to dump $millions into R&D to bring out each new version of their OS.

hmmmmmm...

largo

Maybe it's ready for YOUR enterprise

but not mine. Not on the desktop, at least.

NONE of the publishers of the CAD tools that we use (Cadence, Mentor) are porting to Linux. Mentor ported one of their product lines (not the one we use) and stopped, Cadence never tried, both cite a 'lack of customer demand'.

When I pointed out to the AE that it's hard to have a demand for a product that doesn't exist, and that WE'D jump to it if offered, he just shrugged. Must be nice to publish a product that you just can't get anywhere else....

Those that can

I work for a rather nice sized ISP in Michigan, and no it's not AOL. Every_single_server we have runs linux. We have one, count 'em one, NT server in the entire place that belongs to us and that is here only to give the owners son a job as an NT Admin, sad huh?
For the last 3 or maybe longer years we have run Debian exclusively. Is Linux ready for the enterprise? Yes. Is Linux ready for every enterprise? No. But those of us who can, do.

I would have to say "no".

Major Security hole in virtually all Linux distributions [cnet.com]

Better late than never...

Of course I posted this days ago and was rejected. Oh well.

64 bit

Linux will not truely be viable for the Enterprise until it is entirely 64 bit, not just the kernel. We (a major university that works with geophysical data sets that are large - like corperate datasets) have to kludge around the 2gb limit emposed by utilities like gzip. Without 64bit apps, a 64bit kernel is off little use.

Which enterprise?

There are different levels of enterprises. Are they talking about billion dollar companies, or are they talking about mom-and-pop companies?

Where in the enterprise do they see Linux running?
Is it as a printserver, database server, or the desktop?

Personally, I don't think Windows will be replaced on the desktop in the foreseeable future. The average ubergeek/Linux user hates the normal user too much for that to happen (personally I think it's an inferiority complex...)

On the server side, at my work, we still haven't seen any major request for Linux solutions (we write custom management solutions for midlevel enterprises, i.e. adding specific monitoring support into HP NNM, CA TNG, Tivoli etc). Here, Solaris reigns supreme (if you don't have Solaris support, you can forget it). We looked at Linux support about 6-12 months ago, and the thing is, from a network management point of view, Linux is terrible (right now). SNMP is not fully supported (the UCDavis agent that comes with RedHat doesn't have full SNMPv2 support), and it's harder to get to the underlying hardware than for WinNT and Solaris.

It is getting there though. The Tivoli agent (no matter how you feel about Tivoli TME) has been ported to Linux (at least it was, last time I talked to a Tivoli rep at LinuxWorld '99). The new management standard, WBEM, seems to get full Linux attention from people like IBM, although it's still not there compared to what Sun and Microsoft has in place (basically, the frontend (cimom) seems to be there, the backend (providers) is missing).

Quite frankly, I would be curious to see how people like Google is managing thousands of Linux servers (they're not going around pinging each server each day to see if it's up, are they?)

Sorry about the rambling, it's early in the morning, and it's slashdot.

Still Does Not Answer My Question

This is all fine and good - using Linux for servers is a great business decision. No licensing hassles, stays up like a champ and keeps on performing. End of story. Let's move on.

But what about:

Doing system administration for large LANs of Linux desktops?

Over the years we've been running RISC workstations that are becoming increasingly expensive from a hardware standpoint relative to what can be got in the x86 world.

We'd like to take advantage of the price performance advantage in hardware as well as the increasing maturity of Linux desktop end user applications (which are getting real close now). It seems like a lot more applications are available for Linux desktop than many of the traditional commercial Unices.

The problem is that everyone I know that runs Linux runs their workstation or laptop as their own cowboy system administrator. They typically don't worry about integrating dozens or hundreds of these things together in such a way that a small support staff can manage them effectively.

You know the kinds of systems.

• Haphazard applications installed whereever they felt like
• distro installed out of the box without enough applications
• no patches applied for necessary security updates
• strange hardware hanging off moldy interfaces,
• never thinking about whether it might be nice to use something like NIS (but I know that's not good enough) or automount,
• never providing regular user file backup,
• deciding whether to put apps on a central server or each desktop, (pros and cons either way)
• how to handle upgrades,
• etc.

So what I want to know is:

Has anyone done this?

How did it work? What should we look out for? What is the advantages and disadvantages? Good tools? Web sites?

Almost There

There are still a few things lacking in Linux distributions to make it fully "enterprise-ready" (I hate that term). For anyone who has truly spent time with other Unices, the following is obvious - Better Package Management! The ability to test out packages without "commiting" them so, if something breaks, you can immediately roll back to the prior state. Yes, I know you can uninstall and reinstall old packages, but it's NOT the same. Use HP-UX for a while and you'll understand - you can install, remove, commit, rollback, and test packages. In a production environment, it is critical that a newly installed patch or program doesn't break existing systems!

As far as the kernel goes, I think Linux is there. I DON'T think Linux is necessarily ready to compete with NT or 2000 (though I give it 18 more months), since it is still lacking quite a few easy to use admin tools (think of the NT print manager or DHCP admin and you'll understand what I mean), but it is coming along.

Mirroring

One of the biggest things missing in Linux that I see is software mirroring of hard drives. Are there any projects out there aimed at bringing that to Linux?

Linux and the Enterprise

I think Linux will see greater adoption in the enterprise. Why? Java and J2EE.

J2EE is the current wiz-bang development platform for enterprise applications. Looking past the hype, it appears to actually provide some good tools and structure. We're going to be seeing a lot more enterprise applications written on top of WebSphere or WebLogic or some other J2EE server platform. It seems to be the way to go if you don't want to be beholden to M$ and .NET.

The thing is, since this is most all pure Java, all you need is a stable OS that supports your database (Oracle, DB2, etc.) and, of course, has a Java VM. Enter Linux.

Is ANY OS ready for the famed "enterprise"?

At my last company we ran a heterogeneous mix of Sun, Tru64 and Linux for the "real" (ie, core-business related apps, it was a Biotech) stuff and then were an MS bitch on the desktop and for all the "supporting infrastructure" - email/PIM, etc - in other words Exchange, etc. Which meant the only thing we had in the DMZ was the webmail thing - running on IIS.

Which of course handily got hacked, giving access to the entire network - databases, fileservers, the source tree - everything. I can just imagine all the "not ready" head-shaking if that was a Linux box.

So anyway, I would just love to see an "Is MS ready for entermprise/prime-time/mainstream?" article. (I just hate that word btw - mainstream - what does that make me? some sort of dirty, stagnant, out of the way puddle or something?)

Its Ready just have to convince managers

I try to push Linux whereever I can. The only reason I did my MCSE for both NT and 2000 was so that I could push Linux on managers with authority.

Linux is more than ready just have to convince Microsoft loving managment of this. Hell the price is right! Every Microsoft Box that I replace with Linux I feel that I am doing the world a service.

Linux is comming to enterprise - from all sides.

Quite often, when talking about "Linux in companies", people concentrate on big-scale employements, "global 500" companies and such.

However, most people (90%?) work in small to medium enterprises with hunderts of "bizcases" online these days, and most of these are form small companies, schools, universities, NPOs... who simply can't afford the cost of commercial software, but can't afford the cost of a full-time linux/unix administrator either.

Link: MandrakeBizcases [mandrakebizcases.com]

Nothing related to Free Software

I find remarkable that the article is absolutely not related to Free Software. Of course you need to be an expert to know what is Free Software and what is not in the applications said to run under GNU/Linux. The random reader will have a hard time to find out. More disturbing is the fact that if you s/Linux/Solaris/ in the whole article, it reads perfectly well. I was under the impression that GNU/Linux was developped in a spirit that is revolutionary. Apparently that spirit was lost somewhere. Imagine a text about democracy where you could s/democracy/monarchy/ and it makes sense. One would quickly conclude that the text is not really about democracy and really talks about monarchy. Likewise the article says Linux everywhere but is really about proprietary software, proprietary solutions and a proprietary state of mind.

Re:What about Internet Explorer

what about vmware [vmware.com] ?

Re:Uhhh - swap storms and umount corruption...

Yeah and don't ship a product that causes me to spend all weekend capturing data to send to IBM to look at... (OS/390 patches).

Yeah, and don't cause me look ALL over your web site for 8 or 9 patches just to get Oracle to be stable on my server (Solaris).

Yeah, and don't force me to try and run applications in ring 3 just to debug Abends (Novell)

Yeah and don't force me to call you to get some undocumented patch that is called a "hot fix" even though I am only running YOUR software on the machine... (Microsoft)

Hmmm seems like no operating system is ready for prime time to me.

I have ran all of the above and they all have their issues. I now run Red Hat Linux 7.1. It runs well for what we do. Do I upgrade to the latest kernel every time? NO! Do I test new kernels? Yes.

For us it comes down to money.

Solaris - Intel version sucks, Sun hardware is expensive and propritary.

Novell - NDS is awesome, but way to expensive. The Novell kernel sucks to run apps. Oracle is dropping them.

Windows 2k - Security is bad, expensive, and vendor tries everything to lock you in to their product. Who wants to be FORCED to upgrade every three years?

IBM - Way too proprietary and expensive

Apple - Version 10.x looks good, but has same issues as Sun.

Steve Michael

Re:Linux on Alpha Time is Now?

The future of Alpha is dubious at best. For linux to win in the enterprise software support is a must. That means open AND closed source software. And surprise surprise - it will run on linux on intel first and foremost. No in-it-for-the-money ISV will port something that runs on linux/intel to linux/insert-esoteric-hardware-here unless there's a sound business case for it. It's hard enough getting them to support linux/intel - forget the other stuff (for now at least)!

This argument also applies to linux on mainframes (yes I've tried it, yes I hung it som the VM had to be restarted, yes I love linux, yes I hate mainframes).