Linux Server Operating System (OS) Technology Consolidation – Survey Results

PinellasCounty

Business Technology Services (BTS)

Linux Server Operating System (OS) Technology Consolidation – Survey Results

Version 2.0

Prepared by BTSEnterprise Technology Services (ETS)

December 2008

Table Of Contents

Overview

External Survey Summary

Internal Survey Summary

Conclusion

Appendix A – External Survey Responses

Appendix B – Internal Survey Responses

Enterprise Technology ServicesPage 1 of 1510/16/2018

Linux Server Operating System (OS) Technology Consolidation – Survey Results

Overview

Despite numerous efforts, the preparers of this document were able to find very little objective qualitativecomparison between the Linux SuSE/SLES and Red Hat Operating Systems. In an effort to fairly compare the technologies, two surveys were conducted:

  • External Survey

With the assistance of John Becker (OMB), we were able to identify seven Florida counties of similar size to PinellasCounty:

  • Broward
  • Jacksonville
  • Hillsborough
  • Miami-Dade
  • Palm Beach
  • Orange
  • Sarasota

With the assistance of Richard Ginski (Security/BTS), we were able to identify three additional government entities:

  • City of Largo
  • City of Tampa
  • TampaInternationalAirport

The survey questions were:

1. Whatserver operating systemsdoes your IT/IS Department support?

2a. Do you supportdatabase serversrunning Linux?

  1. If so, what versions of Linux?

c. Any issues associated with running DB servers on Linux?

3a.Do you support middleware serversrunning Linux?

  1. If so, what versions of Linux?

c. Any issues associated with runningmiddleware servers on Linux?

4.For eachLinux platform, how many servers are maintained?

5a. Have you prepared any studies comparing your Linux platforms?

  1. If so, would you be willing to share your findings with us?

6. If you support multiple versions of Linux for serversandwantedto consolidate

technologies, which technology would you choose? Why?

  • Internal Survey

We identified the following BTS internal stakeholder groups impacted by the Linux OS technology consolidation:

  • DBA Group
  • Intel Group
  • Linux Group
  • Middleware Group
  • Network Group
  • Security Group
  • Storage Group

The survey questions were:

  1. Are there any distinct advantages associated with supporting a server running SuSE/SLES vs. Red Hat?
  2. Are there any distinct disadvantages associated with supporting a server running SuSE/SLES vs. Red Hat?
  3. Are there any other points associated with either technology that should be considered in this evaluation?

Enterprise Technology ServicesPage 1 of 1510/16/2018

Linux Server Operating System (OS) Technology Consolidation – Survey Results

External Survey Summary

For the External Summary details see Appendix A.

ORGANIZATION / # OF SERVERS RUNNING / CONSOLIDATION
RHEL 3 / RHEL
4 / RHEL 5 / RHEL release not specified / SLES 9 / SLES 10 / SLES release not specified / OTHER / CHOICE / WHY?
Broward – 1 / 2 / n/a
Broward – 2
Jacksonville
Hillsborough – 1 / 50 / 20 / Some / Red Hat /
  • In-house expertise
  • Good Support

Hillsborough – 1 / 45 / 5 / Red Hat /
  • Support
  • Compatibility
  • SuSE lags in compatibility, pricing, and support

Miami-Dade – 1
Miami-Dade – 2 / 61 / n/a
Palm Beach
Orange – 1 / n/a
Orange – 2 / 6 / n/a
Sarasota / 8 / 2 / 1 / n/a
City of Largo / 5 / 7 / n/a
City of Tampa / 4 / 10 / SuSE / Current support structure
TampaInternationalAirport / some / n/a

= No Response

Enterprise Technology ServicesPage 1 of 1510/16/2018

Linux Server Operating System (OS) Technology Consolidation – Survey Results

Internal Survey Summary

For the External Summary details see Appendix B.

STAKEHOLDER / SuSE/SLES / RED HAT
DBA / - Poor Level of Service / + Oracle Support for Red Hat
- Poor Timeliness of Resolutions
Affected IT Project Schedules / + Fewer DBA SUSE/SLES Server (4) to be rebuilt vs. Red
Hat Servers (13)
LINUX
MIDDLEWARE
INTEL / + Supports OES
+ Leverage Novell investment from other
products / + More online support forums
+ Vendor Support more familiar with Red Hat
+ More vendors competent supporting Red Hat
+ More training options
NETWORK / + Currently there is a network application that is not
supported by SuSE/SLES and is supported by an old
version of Red Hat
SECURITY / + Purchase Price / + Superior Support
+ Better Patching
Longer wait time for patches = less secure
+More Solutions
ISV’s support the more popular platforms
More COTS solutions support Red Hat
+Lower COTS Pricing, Higher Quality
Competition between ISV’s drives price down
+Lower Labor Costs
Ease of use and better support = lower BTS labor costs
STORAGE / + Purchase Price / + Multipath easier to configure
+ SAN boot easy installation
+ iSCSI Support
iSCSI is more cost effective than fiber channel. There is the potential to save money.

Conclusion

The External Survey feedback was limited. Of the parties polled, ten responded (representing five of seven counties).

  • Four of the five responding county agencies employ Linux servers.
  • Only one county had consolidated to one version of the Linux OS (Red Hat), citing support, in-house expertise and compatibility as the reasons behind the decision.
  • Only three counties polled employed a significant number (50 +) of Linux servers.

The Internal Survey indicated a strong preference for the Red Hat Linux OS.

Appendix A – External Survey Responses

DETAILS

1. Whatserver operating systemsdoes your IT/IS Department support? / 2a. Do you supportdatabase serversrunning Linux?
B.If so, what versions of Linux?
  1. Any issues associated with running DB servers on Linux?
/ 3a.Do you support middleware serversrunning Linux?
B.If so, what versions of Linux?
  1. Any issues associated with runningmiddleware servers on Linux?
/ 4.For eachLinux platform, how many servers are maintained? / 5a. Have you prepared any studies comparing your Linux platforms?
B. If so, would you be willing to share your findings with us? / 6. If you support multiple versions of Linux for serversandwantedto consolidate
technologies, which technology would you choose? Why?
PARTICIPANT / 1 / 2 / 3 / 4 / 5 / 6
Broward
Harper Delbert / Sun Solaris 9 & 10
IBM AIX 5.2 & 5.3
Redhat Linux ES Rel 4
Windows 2000/2003
Zos V1.7 / Yes, Oracle 10G / No / Two – one for Development/Testing and the second runs the Production images / No / I would choose Red Hat Linux. Red Hat has been in the forefront with technology, hardware, software vendors and applications. Their support is very good and I believe they will be the leader for years to come. Red Hat is still open source and they are only one of a few that have support like the big UNIX operatingsystems (Solaris, AIX, and HP).
Red Hat Enterprise Linux ES Release 4 / n/a / n/a
None at this time / n/a
Hillsborough
Michael C. Hogg / HP-UX
Windows Server 2003
Red Hat 3,4,5
AIX
Suse (going away) / Yes, Red Hat / Oracle and MySQL / We are looking at JBoss. We use other Red Hat products – GFS (Global File System) and Satellite server / 70+ servers running Red Hat:- ES3 - 50, ES4 - 20, a few migrated to version 5 / No. But Suse has been more difficult to support. We require support so Slackware, CentOS etc. is not considered / Red Hat – In house expertise and very good support.
70+ servers running some version of Red Hat, mostly ES but a few AS. Our Suse boxes are being migrated to Red Hat / Red Hat 3,4,5 AS and ES / n/a
They’re more stable than Windows. The biggest issue keeping the in-house Linux gurus and strong DBAs. / (no answer)
Hillsborough
Bruce Dangremond / Windows 2003
Red Hat Linux 4, 5 HP-Unix 11 / Yes we run Oracle 10g on RH Linux / yes / 45 RH 4 servers, 5 RH 5 / No we have standardized on Red Hat. / We have standardized on Red Hat as it is the leader in support and compatibility. SuSE seems to lag in compatibility and dealing with Novell for pricing, maintenance and support should be carefully reviewed before proceeding.
RH Linux 4 / RH 4 / N/A
None that we have experienced, in fact it was the recommended solution from Oracle. / None
Miami-Dade
Peter Oelkers / z/OS 1.9
z/VM 5.2, 5.3
zLinux SUSE Sles 8, 9 & 10
AIX 5.3, 6.1
Solaris 8, 9
Wintel Linux
Various Levels of Windows are supported / Oracle 10g / Only for one application which is our Enterprise Asset Management System which supports Apache and a Databridge function. The majority of our middleware software resides on the pSeries platform utilizing AIX 5.3 / Oracle GRID Production18 (5 GRIDS supporting 60 Databases)
Oracle GRID TEST 9 (3 GRIDS supporting 75 Databases)
EAMS Prod8 (Virtualized with VMWARE)
EAMS Dev8 (Virtualized with VMWARE)
EAMS Staging8 (Virtualized with VMWARE)
zLinux10 / Presently, we are evaluating the migration of workload from AIX to zLinux / At this point we do not have any solid plans to consolidate. The results from question 5a may indicate that consolidation is warranted.
Oracle Enterprise Linux 4.6. / Redhat Enterprise Linux 5.2 / Surely
None to my knowledge / No issues have been found if validation with middleware vendor prior to deploying middleware if software is certified for the specific release of the LINUX OS.
Orange
Chiliano Heredia / Win2k3 R1, R2
IBM AIX 3.5
Z OS 1.7, 1.9 / No not at this time, but we are in the process of performing a Proof of concept in running Oracle on z Linux. / Not at this time / n/a / Referred to answer 2a. / n/a
Has not been decided at this time, probably Red Hat Linux. / n/a / We would be happy to share our findings when we finish our proof of concept.
n/a / n/a
Orange
Jorge Collado / Windows Server 2003 (x86 & x86-64), Windows Server 2008
Red Hat Enterprise Linux 4.6 / Oracle on Linux, configured in RAC environment, and 2 production Oracle standby database servers running
Independently. / Not currently / RHEL4 – Two production database servers (Oracle RAC cluster), two standby database servers and two development database servers. / We are currently only using one Linux distribution / N/A
Red Hat Enterprise Linux 4.6 / N/A / N/A
Our experience has been mostly favorable. We saw a dramatic performance improvement when we switched these databases from Windows 2003 to RHEL.
We have run into a few bugs that we never encountered when running under Windows 2003 (a legacy application required a new version of the Oracle client after we switched to Linux, even though the DB version was the same; and we also encountered an Oracle bug that prevented some records from returning on some queries.) However, we feel that the benefits have far outweighed the issues we have experienced. / N/A
Sarasota
Glenn Zimmerman / Windows NT, 2000, 2003, 2003 R2
SuSE Linux v10
SuSE Enterprise Linux 9
Oracle Enterprise Linux 5
AIX 4 / Yes, Oracle 9i; Oracle 10g / Our perimeter mail servers / 8 SuSE Enterprise Linux 9
2 SuSE Linux 10
1 Oracle Enterprise Linux 5 / No / No determination has been made
SuSE Enterprise Linux 9
Oracle Enterprise Linux 5 / SuSE Linux 10
None / None
City of Largo / RHEL 3,4 and 5
OpenSUSE 10.2, 10.3, 11
SUSE-SLED
SUSE-SLES / Yes / No / RHEL 5
SUSE 2
OpenSUSE 5
(Probably some I missed along with numerous test servers) / No / Becauseof the potential for support issues, the product or application youwant to run should always dictate what Linux OS you run.
RHEL 3,4 and 5 / n/a / n/a
No / n/a
City of Tampa / SUSE SLED
Red Hat / Yes / Yes / SUSE 10
Red Hat 3,4 / None, business position and applications drove the move (much like
it did for Windows) / SUSE due to current support structure
MySQL / Red Hat / n/a
No / No
TampaInternationalAirport / Windows NT
Windows 2000 Windows 2003
Linux Redhat / Yes / Yes / 3 database
3 application / No
Red Hat ES 4 / Red Hat ES 4 / n/a
No / No

Enterprise Technology ServicesPage 1 of 1510/16/2018

Linux Server Operating System (OS) Technology Consolidation – Survey Results

Appendix B – Internal Survey Responses

Respondent / RESPONSE
DBA /
  • Are there any distinct advantages associated with supporting a server running SuSE/SLES vs. Red Hat?
ADVANTAGE FOR Red Hat FOR ORACLE DATABASE SERVERS: Oracle Corporation’s expertise is in Red Hat. Oracle has staff that knows the internals of Red Hat because it offers its own flavor of Red Hat.
  • Are there any distinct disadvantages associated with supporting a server running SuSE/SLES vs. Red Hat?
HISTORICAL DISADVANTAGE FOR SuSE/SLES: in 2007 the linux server staff experienced issues with the level of service and timeliness of resolution from Novell. Please refer to Jeff Scarsbrook’s document “RHEL over SuSe_08072007.doc”
HISTORICAL DISADVANTAGE FOR SuSE/SLES: In 2007, server/dba staff reached a roadblock in that Sles 10 had issues thatwent unresolved for months that prevented Oracle 10g database servers from being built out with Sles 10. i.e. project schedules could not be met. Please refer to Jeff Scarsbrook’s document “RHEL over SuSe_08072007.doc”
  • Are there any other points associated with either technology that should be considered in this evaluation?
AMOUNT OF TIME FOR SERVERS TO BE REBUILD BY BTS DBA AND LINUX GROUPS DEPENDING ON WHAT THE FINAL STANDARD IS:
-If Red Hat is selected 4 SuSE/SLES database servers will need to be rebuilt.
-If SuSE/SLES is selected is 13 Red Hat database servers will need to be rebuilt.
(see OS_Count_TS0001_08122008.xls)
Linux
Middleware
INTEL /
  • Are there any distinct advantages associated with supporting a server running SuSE/SLES vs. Red Hat?
None other that OES services from Novell running on SLES.
Leverage Novell investment from other products
  • Are there any distinct disadvantages associated with supporting a server running SuSE/SLES vs. Red Hat?
There are many more forums and support sites online that support Red Hat. It is often easier to get feedback from user groups. In this country most technology companies provide support for Red Hat Linux before supporting SUSE/SLES Linux. The technical support personnel from the vendor are generally more familiar with Red Hat Linux.
Red Hat is a market leader and because of this has broader 3rd party support and more training options
Our staff has had more success and a generally better opinion of Red Hat
  • Are there any other points associated with either technology that should be considered in this evaluation?
What are our current vendors most comfortable supporting? I believe in most cases, with Novell being the exception, our vendors are more competent supporting Red Hat.
Cost of time for migration of current standards to proposed new standards.
True cost of support, IE if SUSE support is less expensive but not responsive is it really a better value?
Network /
  • Are there any distinct advantages associated with supporting a server running SuSE/SLES vs. Red Hat?
From a Network & Telecommunications perspective, there is no real difference in supporting either OS. We do have a server that uses an older version of RH due to vendor support for RH only and at a older version.
  • Are there any distinct disadvantages associated with supporting a server running SuSE/SLES vs. Red Hat?
From a Network & Telecommunications perspective, there is no real difference in supporting either OS.
  • Are there any other points associated with either technology that should be considered in this evaluation?
None.
Security /
  • Are there any distinct advantages associated with supporting a server running SuSE/SLES vs. Red Hat?
The only one I can think of is that transition costs to another platform would be avoided. But transition costs are “fixed costs” which, like licensing costs, “evaporate” over time.
Licensing costs appear to be less on the SUSE side.
But the support issues/problems and labor issues (variable costs that accrue indefinitely over time) with SLEZ would continue. Over time, this factor would outweigh initial fixed costs (licensing and transition).
  • Are there any distinct disadvantages associated with supporting a server running SuSE/SLES vs. Red Hat?
Yes!
1)Better Patching
Patching on the current platform will continue to have more of a delayed-affect and the County will be more exposed/less-secure, due to proprietary elements in the SLEZ distro. We should be able to take patches made-available by the Linux community and install them directly on our Linux systems without having to wait on vender to tweak it so that it will work on their distro. If a vendor (Novell) has to take these patches from the Linux community and then covert them to their SLEZ distro so that it can be installed on SLEZ, it will increase our length of exposure making us less secure. For example, current distros of SLEZ do not contain the latest open source code from the Linux community. More specifically, I know that SNORT NIDS, included with SLEZ, is at least two versions behind what is available by the community. Not good.
2)More Solutions - Better Support
Our department strategy is to re-use, buy, then build as a last resort. This essentially means that the bulk of new technologies/new solutions will be purchased if we don’t already have them. SUSE and Novell have long been a niche market in the US which comes with disadvantages for them and their customers like us. We have been held back and our hands have been tied by staying with them
More Solutions:
ISV’s support platforms that give them the best opportunity of making revenue. Therefore, from personal experience, ISV’s support the more popular platforms first in order to maximize revenue. Then, they later “port” to other distros. Their solutions, which are ported-later to other distro’s, tend to be less reliable because there is less focus/effort on those other distros. This negatively-impacts our agility. The other distro’s that are later supported are almost treated as an after-thought by ISV’s which results in support issues on these niche platforms from ISV’s.
Further, ISVs’ support agreements typically-don’t support Linux as a whole. ISV’s typically will only support specific distros of Linux.
More COTS Solutions:
Given all of this, due to the strength and market presence of Red Hat (which includes RHEL, CentOs, and Fedora), has resulted in more of the mainstream ISV’s (Independent Software vendors who provide COTS solutions) supporting Red Hat than any other distro. This results in more solutions being made available on Red Hat platforms and therefore, more-diverse solutions that BTS can offer to their customers. This would place us in a stronger competitive position. When, in a number of cases, only one solution from one vendor is available for a given distribution, this places BTS and their customers in a weak and undesirable position. Finally, the customer essentially looks elsewhere because BTS limited what solutions they can provide their customer.
3)Lower COTS Pricing, Higher Quality
Further, if using more-popular distros, there is more competition among ISV’s. “Economics” dictates that more competition results in lower ISV pricing (for COTS) and higher-quality. This results BOTH the customer and BTS “winning” in these scenarios due to lower pricing and higher quality solutions. In the past, we have not been able to take advantage of this scenario because we have been involved with “niche-players”.
4)Lower Labor Costs:
This needs more analysis. But I believe the key argument in this whole issue is labor cost. Unfortunately, ETS has not been tracking labors hours spent on SUSE vs. Red Hat. Without gathering this vital information, cost analysis, in general, can not be determined. You can’t just compare licensing costs here. Licensing costs are fixed costs and fixed costs “vaporize” over time. The labor costs are significant variable costs that will continue to exist as long as you have the technology. Bottom line, fixed costs eventually “go away” over time, variable costs remain with you. So variable costs are really where the focus should be.
  • Are there any other points associated with either technology that should be considered in this evaluation?
Service Provider of “Little Choice”
Whether is has been Novell’s OS, SUSE, or other products being offered by Novell, there have been many times I have heard the statement “we’re waiting on Novell” to either resolve a problem or to provide functionality that other solutions already have in production (and Novell claims is in theirs but does not work). This has been occurring since I have been working here (11 years or so).
Given this, there have been indications that Novell struggles with sustainability. For example, they have released products which are incomplete, example: Novell’s certificate authority where a component critical to its functionality was not even included (CRL’s, certificate revocation lists in order to disable certificates that are distributed). Or when they released NAM with a “logic bomb” still included in their “production product” where NAM died on us, due to the logic bomb. This not only negatively-impacts our service offerings but increases costs and damages our reputation.
Although the company has been around for decades, they still operate very much like an immature company.
They also continue to struggle financially by incurring more losses than they do profits. For BTS, this is not a good indicator of long-term sustainability (one of our principles). Granted, their Balance Sheet may show Novell’s “liquidity”, but that is not an indicator of profit/loss. (Their Income Statements indicate profit/loss). Further, mathematically speaking, Novell’s liquidity will disappear if they continue to sustain losses.
In order for us to be the “Provider of Choice” we need to use “Vendors of Choice”. At some point, as the BTS organization matures, BTS will be only as good as the vendors they have chosen to work with. If we have to compete with other providers who are using “better vendors” (who are providing more/better solutions) we will lose “in the face of competition”.
Comparison Over Time:
Instead of a 3 year comparison, we need to realistically consider how long we will be leveraging this technology once the decision is made to go in a certain direction. For example, most of our platforms have been around for 10+ years. I would think a similar length of time would need to be utilized when comparing the two options and their relevant costs over time (10+ years).
Storage /
  • Are there any distinct advantages associated with supporting a server running SuSE/SLES vs. Red Hat?
Yes, the SuSE/SLES software license is cheaper.
  • Are there any distinct disadvantages associated with supporting a server running SuSE/SLES vs. Red Hat?
Yes, SuSE’s multipath and san boot installation isn’t built into the native SLES install (Requires extra configuration)
If a San volume is added to the server, it has to be rebooted.
No iSCSI support
  • Are there any other points associated with either technology that should be considered in this evaluation?
Red Hat incorporated Multipathing and San booting in the native Red Hat installation
Red Hat has iSCSI support
If a San volume is added to a Red Hat server it does not require a reboot.

Enterprise Technology ServicesPage 1 of 1510/16/2018