Category: guardtime

Posted on

rsyslog’s first signature provider: why Guardtime?

The new has already spread: rsyslog 7.3 is the first version that natively supports log signatures, and does so via a newly introduced open signature provider interface. A lot of my followers quickly realized that and begun to play with it. To make sense of the provider interface, one obviously also needs a signature provider. I selected the keyless signature infrastructure (KSI), which is being engineered by the OpenKSI group. Quickly, I was asked what were the compelling reasons to provide the first signature provider for this technology.

So rather than writing many mails, I thought I blog about the reason ;) Continue reading “rsyslog’s first signature provider: why Guardtime?”

Posted on

Joining the Guardtime Technical Advisory Board

I just wanted to let everyone know that I will be joining the Guardtime technical advisory board. The board’s prime mission is to make sure that Guardtime implements technology that users need.

While implementing the rsyslog log signature interface and it’s first signature provider, I worked very closely and productive with the Guardtime folks. Over time, we exchanged more and more ideas and the whole thing culminated when I visited their office in Tallinn, Estonia where we had even better discussions and hacking sessions.

Guardtime seems to have enjoyed this work as well, and so I was recently asked if I would like to join their to-be-formed technical advisory board (TAB). At first, I was a bit hesitant because I wondered if there is any collision of interest between being on the Guardtime  TAB and driving rsyslog forward. So I talked a bit with the Guardtime folks clarifying my position. As it turns out, the whole point of the TAB is to assemble experts from different fields and let them discuss and provide information on what end-users really need. The rationale is very obvious: Guardtime as a company can only succeed if they provide what the market really needs. There is no point in trying the market to pick up some product or technology that has no real value. Too many companies tried that in the past and failed miserably. So, in that sense, I can act as the logging user’s agent inside the TAB and help push Guardtime in a direction that really is of value for logging. As I already think their current technology is a really good fit for what we need in logging, this is an exciting opportunity.

I also admit that it is very appealing that I will get ready access to Guardtime’s team of qualified cryptographers (including chief scientist Atho Buldas, who helped building the Estonian national PKI), which is a really big win in regard to secure cryptography in rsyslog. I never before had the chance to work with such top-notch crypto people. In fact, they already provided feedback on the non-Guardtime encryption bits in rsyslog and I am sure that working closer with them will help us getting very solid cryptography into rsyslog. If in doubt, just look at the rsyslog git tree and commit history about when all this new crypto stuff got in :-)

As a side-note, I am also talking with some other party about a custom module using a HMAC based tamper proof system for log records. While this is not as strong as the Gurdtime signature provider, there are obviously some use cases for it as well. If that project materializes, it will of course be merged into the main rsyslog git tree. So, for the concerned, rsyslog will definitely not get a Guardtime bias. After all, my main interest is driving rsyslog forward and, again, you can do that against what “the market” demands.

One of the things I discussed with the Guardtime cryptographers while I was in Tallinn was about how we could extend log integrity proof to a full network relay case, where relays even drop messages. This is a very hard and yet unsolved problem (for details see my paper on log signing in the relay case). While we got no ready solution from the initial discussion, we had some very promising ideas. This for sure is nothing we’ll see within the next quarter or so, but I am happy I now have the opportunity to discuss this and other hard problems with the knowledgeable (and very friendly, btw ;)) folks from Guardtime.

So for me getting on the TAB I think is a win-win-win situation for rsyslog, it’s user base and of course also Guardtime. Hope a lot of good stuff will come out of it.

Posted on

What is a rsyslog signature provider?

Some folks asked what a (rsyslog) signature provider is. In essence, it is just a piece of software written to a specific interface.

There are little functional requirements for signature providers. Most obviously, we expect that it will provide some useful way to sign logs, what means it should be able to provide an integrity proof of either a complete log file or even some log records some time after the logs are recorded.

There is no “official signature provider spec” available. As usual in open source, the provider interface is well defined inside its actual source header file. Take a look for example at rsyslog’s git to see the definition.

If you look at that code, the interface is fairly simple. The primary entry points a programmer needs to define are

  • OnFileOpen
  • OnRecordWrite
  • OnFileClose

As the names imply, they are called whenever a log file is opened, written to, and closed. Again, what the provider does at these events is up to the provider.

The initial provider that ships directly with rsyslog is using a secure remote signing service provided by Guardtime.  There are various reasons why I selected this provider, and I elaborated at length on them in my blog posting on why I selected Guardtime as first rsyslog signature provider. In a nutshell, it is a method that does not rely on local secrets and as such can not be circumvented (again: details in my other posting). Obviously, Guardtime offers paid services, but they also operate free services for everyone to use (a key point when I evaluated that technology).

If you have privacy concerns about the Guardtime provider, you may want to read my blog posting on just that question.

Note that signature provider implementations can use the above entry points in any way they like. For example, the may ignore file open and close and do all work in record write. So this small interface is highly flexible. If you look at the full interface description, you’ll also see some more events which primarily provide housekeeping functionality. If you intend to write your own signature provider, I suggest to start with a copy of the guardtime provider and change it according to your needs. That way, you have a great example of how the housekeeping calls are handled.