The other day I was reading a blog post on handling syslog at scale back on cribl.io’s blog. As you can imagine, syslog-ng has been used to solve syslog related challenges for a while now (24 years to be exact) and with that expertise I wanted to point out a few things in relation to that blog post. This might even become a series.
The blog post linked above, gives some advice how to scale syslog, in the section titled: Scaling Syslog the Right Way. Read the blog post for more details, but here’s my summary:
place the receiver (e.g. Cribl Stream) right next to your log sources (e.g. in the same data center)
scale out the receiver over many nodes so it becomes a cluster
make sure to deploy a load balancer in front
What this means in practice is that you will need a sizeable infrastructure to consume the logs of a syslog producing device. Since my assumption is that all data centers have such appliances, you will need to deploy this infrastructure in each of your data centres (to be close to the sources).
While I can see that load balancing clusters to process log data are important in some scenarios, I don’t think this use case should be one of them. A single node, potentially in a failover High Availability setup should suffice.
There’s a choice between scaling out vs scaling up a workload. Cribl recommends scaling it out. My take is that it should be possible to scale it up before scaling it out.
syslog-ng has a bag of tricks to make it fast even on a single node, thereby reducing hardware costs and operational complexities and the need for a load balancer.
it is implemented in C (compared to Cribl’s choice of TypeScript, fluentd and logstash are Ruby IIRC)
it avoids/minimizes copying of data while processing them, its log routing implementation uses copy-on-write semantics for cases where multiple paths potentially change the message in parallel paths
it avoids/minimizes memory allocations, in the simplest case it allocates 1 block of memory for 1 message
it uses an efficient asynchronous architecture, using epoll and one thread per CPU core
it uses a message representation where fields (aka: name-value pairs or properties) are stored in a packed block of memory, that is efficient to look up & serialize.
it uses internal queueing mechanisms that avoids lock contention and allows back-pressure to be applied selectively
it offers alternatives to regular expressions, as regexps are pretty slow to evaluate at volume
I understand that this is an apples to oranges comparison. Cribl seems to have good UX. syslog-ng has good performance.
This is on my ~2018 laptop (Intel(R) Core(TM) i5-7440HQ CPU @ 2.80GHz, 4 cores), with a single destination writing all messages to disk, with syslog parsing enabled.
In my last post, I enumerated the long term objectives I distilled from the discussions I had earlier with some syslog-ng users. Thanks for everyone who responded to that and/or filled out the survey, very insightful responses, something to work from. The survey is still open, but as always, the more the better so some more responses would be very much appreciated.
You might be operating a logging infrastructure, or you might be in the process of deploying one, it does not matter: if your job or project involves logging and solving the problems it entails I’d like to know your opinion.
I find that a lot of syslog-ng deployments are lagging behind and are using ancient versions. It has become difficult for me to get these deployments to more recent versions. No product is able to improve and cover new ground in a situation like this…
Maintaining this change velocity in the log management space is not feasible. Deploying a log management and processing infrastructure from scratch can literally take years just one time. Swapping out technologies every now and then on a whim would mean that the project never reaches the goals it was set out to achieve.
With that I said I still think that being able to regularly push out updates to deployments is an important bottleneck to solve for any product to be sustainable. This is needed for both the feature front (e.g. addressing new use-cases) and on the support perspective (e.g. fixing bugs).
I often get questions about syslog-ng 3.5.6. This release was originally published 5th August 2014, roughly 8 years ago, and happens to be part of EPEL7. syslog-ng is included in BMW i3 vehicles, this video shows the listing of open source components on the infotainment screen, The BMW Open Source DVD contains syslog-ng 3.4.7, a whooping fresh release from December 2013. There are similar stories with syslog-ng included in products or an OS release, usually with pretty old versions.
Why does this happen?
Due to the early adoption of syslog-ng, it was included in a number of Linux distributions and BSDs/UNIXes, even became default in some of them. I considered this a great success.
For none of these distributions however is log management a central question. They each need some kind of log daemon, but that’s it. Whether that log daemon is syslogd, rsyslog or syslog-ng does not really matter. Neither matters their actual version number. So even though distributions helped initial syslog-ng adoption, they have become a bottleneck in delivering new releases to users.
Users can still upgrade, right?
Enterprise users (and products that embed Linux and syslog-ng) pick an OS version and plan with it for ~10 years. Unfortunately they deploy syslog-ng as a part of the OS and expect the OS vendor to provide support. Often, the sysadmin responsible for log management is not even allowed to upgrade. Some claim that upgrading syslog-ng would violate their support terms, causing the entire OS to become unsupported.
So even though more recent versions of syslog-ng includes functionality or fixes they need, they stick to the old version and try to work around any issues they find.
The support from the OS vendor for the logging component is questionable at best and is restricted only for the most basic use-cases, not cases where syslog-ng would play an important role in one’s infrastructure. Just as log management is not a central focus for the OS, neither is it for the support team behind the OS. They would fix security issues, should they be reported, but otherwise they will just continue to use what they have.
Solution: state of the art binaries to pick from
Building the latest version of syslog-ng for your enterprise distro on your own is not for the faint of heart. Even though 20 years ago, building your own kernel or application was an essential part of a sysadmin’s job on any UNIX, this is not true any more.
Also, it was a lot easier to build syslog-ng in 2001, today we have so many integrations that pulling all the build dependencies (and the right versions) is far from trivial.
We worked hard in the past years to resolve this issue and today syslog-ng is not only available in source format. There are a number of options today to pick from, should you want to use the latest and greatest:
Over time, the building of bespoke/customized packages has become much easier too, this blog post explains it all.
So what’s your excuse? I am really interested if the options above suffice. Do you still use an old syslog-ng version? Why? Would any of the above work for you? If not, What would YOU need to upgrade syslog-ng to recent versions? And what would you need to change your processes to plan for upgrades regularly?
If you have a response to any of these questions, please post it as comment below or drop me an email. Thanks.
I mentioned in the previous post that I would like to focus on syslog-ng and put it more into the spotlight. I also mentioned that Balabit, the company I was a founder of and the commercial sponsor behind syslog-ng, was acquired by One Identity ~4 years ago. How does this add up? Who owns the Intellectual Property (IP) for syslog-ng? Who am I or this blog affiliated with?
I felt this post was important to set things straight and make it easier to understand my motivation. If you are not much into Free Software and Open Source licenses or not interested too much in administrative nuisances of FLOSS projects, feel free to skip this post.
First of all, the IP in syslog-ng that Balabit owned originally, was transferred to the acquirer, One Identity. This includes:
copyright on the documentation and parts of the codebase
With this change we’ve stopped requiring signed CLAs (Contributor’s License Agreements) whenever someone contributed to syslog-ng. This means that the copyright of any outside contributions would be retained by the contributor and not assigned to Balabit or its successors.
Over the years, many such outside contributions were merged into the syslog-ng codebase, meaning that the code today is owned by many different individuals and companies. Copyrights of files are tracked by the tests/copyright/policy file in the source tree: you can note that there are some files that are external contributions in their entirety. Some files have mixed copyrights, partly owned by One Identity, partly by the outside contributor.
Since the license syslog-ng uses is a combination of GPL + LGPL, the combined work as such is free software forever. The GPL/LGPL warrants that anyone can get the source code and be able to change it in any way he or she wishes. The sole requirement of the license is that should you distribute syslog-ng to any 3rd parties, you would need to disclose that you were using GPLed code and offer the source code along with any changes you have made.
The same rules apply to the commercial sponsor as well: as long as it relies on the work that was produced by external contributors over the years, they will need to publish any changes to syslog-ng they create. The only way out would be to rip out all code created by external contributions.
This means that syslog-ng today is a truly open source project, at least from the licensing perspective
But there’s another perspective, namely whether there is an active developer community that adds new features and publishes new releases.
Fortunately, this also holds true. The syslog-ng project lives on https://github.com/syslog-ng/syslog-ng, contributions are transparently managed in GitHub pull requests, either if created by One Identity or someone else in the community. Regular releases are produced on a 8-10 weeks cadence and are published both in source + binary formats and a docker image.
With all of the explanations above, the status of this site and myself can easily be explained: I am not affiliated with One Identity (the successor of Balabit) in any way. I am an individual who contributes time and energy to the open source syslog-ng project, just as I did the same in the last ~24 years.
This is a personal blog, related to syslog-ng and producing useful fixes and features for syslog-ng as an open source project. I am not sponsored or endorsed by One Identity. I am here to help finding out where syslog-ng should go next.
The consensus of the FLOSS community is that a project is only considered truly open source if the developer base/IP is not concentrated within a single company/organization. The reason for this understanding is simple: if there’s only one such entity then that entity has too much control/power over the project. If the company goes bankrupt or changes hands, then priorities might change in a way that causes the open source project to suffer. The long term sustainability of an open source project hinges on the breadth of its contributor base: the broader it is, the more likely it is that the open source project can outlive its creators or commercial sponsors.
The dependence of the syslog-ng project on Balabit as its commercial sponsor has been an issue since ~2009 and probably the reason why it has not become the default logging daemon in Fedora, thus RedHat Enterprise Linux. I don’t think the inclusion in Linux distributions is the prime venue of competition today as it once was. But this decision by Fedora still hurts. 🙂
In a way, the acquisition of Balabit, and my departure from One Identity later allows syslog-ng to become a truly independent-, sustainable open source project.
syslog-ng has been around for decades: I started coding the first version of syslog-ng in September 1998, circa 24 years ago. The adoption of syslog-ng skyrocketed soon after that: people installed it in place of the traditional syslogd across the globe. It was packaged for Debian, Gentoo, SUSE and even commercial UNIXes. It became a default logging daemon in some of these Linux distributions. Commercial products started embedding it as a system component. Over the years however I feel that syslog-ng has become a trusted piece of infrastructure, few people really care about. I set out to change that.
The use of syslog-ng has become so widespread and dominant, needing minimal maintenance, that after a point, people stopped noticing its existence. It became like the printer sitting in an office corner: we know it’s there, we use it regularly, we appreciate the function but we don’t really know or care about the details or the brand providing us with given service.
I see syslog-ng regularly in this spot today: its deployment might have been a big project in its time with its own challenges, but it has been a solved problem ever since.
Not that log management and log processing would be a static, boring field of IT & IT Security. Like all other fields of enterprise IT, there’s been tremendous activity in the last 10-15 years.
Markets and relevant trends:
SIEM & User Behavior Analytics(LogLogic, ArcSight, QRadar, Splunk, …)
Big Data (Hadoop, Kafka, Storm, Spark, NiFi)
Enterprise SaaS services (Office365, Google Workspace, etc.)
Containers and orchestration (Kubernetes, OpenShift, cloud & on-prem)
Cloud Native Applications
All these changes naturally resulted in an equal frenzy in the tools processing and managing log data. New tools and services emerged, old tools gained new features. I could probably go on and get into details on these trends but that’s not why I am here today.
I started this blog as I wanted to show two things:
That syslog-ng has not been the stoic figure in the corner and has incorporated important improvements over the years that are not widely known and unfortunately not even assumed.
To solicit feedback on my future plans and with that help guide the development of syslog-ng to the future.
The intent behind this blog is to address the 2nd point.
The first point might sound a little strange at first: if there are indeed functionality in syslog-ng that its users don’t know or care about, that can only mean one of two things:
Those features were not needed in the first place.
The marketing/communication of syslog-ng as a project has not been very good.
As one of the engineers behind the changes I firmly believe #1 is not true. The features we added to syslog-ng over the years are important. I believe these features enable syslog-ng to address problems that only few people assume it could address. But I am not here to go into details on those features either.
My take on the marketing issue is different: other projects, open source or commercial, have been better at communicating their value propositions. They were more successful at communicating their release-by-release improvements and with that gained a more significant traction in the marketplace.
The reason behind this failure is an entire post on its own (let me know if you are interested!), my short and simple summary is a single word: focus.
I am the founder of the syslog-ng project. I founded a company that sponsored the syslog-ng project. But neither my or my company’s primary focus has ever been syslog-ng. Some of you may remember that syslog-ng was hosted on balabit.com. Balabit was a player in the Privileged Access Management space (e.g. the likes of CyberArk, BeyondTrust, e-DMZ, Wallix etc). Albeit we made an effort to combine log management with PAM, but truth be told we never really succeeded in doing so. syslog-ng grew from being my personal hobby to become the 2nd product in the Balabit portfolio.
This situation handicapped syslog-ng compared to those projects and companies that had logs as their primary focus.
Balabit was acquired 4 years ago: I spent my sabbatical, I learnt a couple of new hobbies (electronics mainly, welding is something I still want to learn), implemented home automation in my house (see http://bazsi.blogspot.com/), became a hobby angel investor and a management consultant. With all that I am somewhat bored. I love spending time with my family all these new things, but at the same time I need new challenges. There are too many “small” things I spend my time with and I have an itch to do something “bigger”.
I want to give syslog-ng a chance it never had: I want to make it my primary focus. The foundations and the technology are already there, let’s put the spotlights on, blow the dust off. Engage with users, understand their needs and communicate value. Understand things that are missing and fix them.
In a nutshell, I would like to relaunch syslog-ng as a project. Let’s reboot the process that keeps a product able to adapt to a changing market and continue to be relevant for more decades to come.
I am inviting you to be a part of it. Feedback, new use cases, feature requests and even bug reports are welcome. Strong points that you like, weak spots that you would like to see improved are very interesting.
Subscribe below and help me in this endeavour. Stay tuned!
Strictly Necessary Cookies
Strictly Necessary Cookie should be enabled at all times so that we can save your preferences for cookie settings.
If you disable this cookie, we will not be able to save your preferences. This means that every time you visit this website you will need to enable or disable cookies again.
3rd Party Cookies
This website uses Google Analytics to collect anonymous information such as the number of visitors to the site, and the most popular pages.
Keeping this cookie enabled helps us to improve our website.
Please enable Strictly Necessary Cookies first so that we can save your preferences!