Hippity hop

Exciting new tool to heal DevOps redundancy pain goes live

Lucy Carey
LAN

JFrog spawns “world’s first” high-availability binary repository manager in flagship product update, Artifactory 3.1.

A
developer survey
carried out by JFrog, the creators of open
source repository manager Artifactory and software distribution
platform Bintray, revealed that 65% of developers are concerned
that their software (OSS) is unreliable. This was backed up by
feedback from their existing customer base, who complained that the
bulk of their releases were delayed by code and tooling issues and
endless integration tests.

Out of these prevalent ‘pain points’ came the
inspiration for JFrog’s latest release – Artifactory 3.1, which
comes complete with what the comapany are billing as the world’s
first  high-availability binary repository manager. This zippy
new solution went live today – and you can click here to
download – but first, we highly recommend reading our interview
with JFrog CTO Yoav Landman for the full story behind this
release:

JAX: Can you walk us through the thought
process behind the development of Artifactory 3.1?

Yoav: Part of what we saw in our
survey last year is that there is a huge pain in managing open
source components and open source software. And another thing that
we got is that, part of managing open source software, and part of
managing binaries and artifacts in general, the management tools –
the repositories that people are using to manage artifacts – are
becoming something that is of production quality.

So, when your artifact manager is down, basically the
whole development production floor is paralysed. No new builds can
go in. Everything today is fully automated, from development to QA,
to production, so if a tool like Artifactory, which is where the
build artifacts end their life in the production system – if it’s
down, the whole production is down – and it’s unacceptable.

That’s why we developed this solution. We’ve been
getting this feedback not only from the survey, but from customers,
that we know go all the way to production with artifactory.

Everytime developers need to update the tool, they
can’t afford the downtime of upgrading. Also, in terms of load on
the tool, there are many, many more consumers. The repository
manager became such a central point of development, where
developers deposit binaries, and then QA takes them for testing,
and finally they put them in production, or they throw them away –
so the loads on the repository manager became so huge that we
required multiple instances a lot of the time to manage the
flow.

So this solution solves two problems: The first is to
distribute the load, and the second, which is we actually got more
demand for, is to make sure that there is no down time for your
artifactory server.

Artifactory 3.1 

What were the biggest challenges that you
faced in making this project come together?

The first one was coming up with a good and simple
architecture. One of the main goals that we set for the theme and
the architecture group that we set up was that configuring and
 installing the software should be very simple.

The second challenge was [making sure] that adding new
servers was also going to be super easy. We wanted it to be that,
once the original server was in place, it would be very simple to
add new servers on to your cluster. Finally, the other challenge
was just making sure that you have good visibility as a manager, as
an administrator, to understand the components, so that it’s very
easy to see the state of the full cluster, if a server is
down…and so on.

Do you think that this release will change how
people view Artifactory?

Basically, since the need came from our current market
– it was a demand for a much more robust artifact management
solution – so the target audience is mainly enterprises – and it’s
mainly enterprises that are already using Artifactory, or are
considering adopting Artifactory, and are holding back because they
think that the homegrown solution may be more resilient. That’s no
longer true.

How has your client base grown in the last
year – have you seen a lot more people coming on board in
general?

Yes, actually. JFrog basically paved the way for a
whole new domain – which is the whole artifact management
business.

We started to see the adoption curve getting much more
velocity over the past year. We are seeing not only early adopters
and startups. We’ve seen large enterprise
adoption before – but we’re starting to see
legacy enterprises adopting Artifactory and using it as a tool for
managing legacy builds. Like C and C++ builds.

Basically, what’s happening in the domain today is
that those builds which were once monolithic builds that used to
take days or even weeks until we could get a version are going
through the phase of adopting continuous integration, building up
those systems into much more modular, smaller builds that allow
them to be more agile and to release more often.

Artifactory is no exception to that. It’s becoming part of
the chain. We see a lot of legacy systems moving to manage their
build artifacts on Artifactory. That’s one big change we saw in the
past year.

What’s in store for JFrog this
year?

The next news will be related to Bintray. Right now, the
whole module system is just exploding. Because of the cloud, and
because of the fact that it doesn’t make
sense for anyone to develop
a usable system, or any kind of software
system, without making it modular, without supporting tagging,
without supporting extensibility to modules that can be downloaded
– those  modules can be technical packages that give your
system more functionality, they can be small applications like
mobile devices – so anything that’s consuming binaries from the
cloud today.

And this is part of the reason why we developed
Bintray. We are seeing a huge growth – even every new programming
language has modules that can be consumed from the cloud – so we
see a huge adoption and a huge growth in downloadable, pluggable
software from the cloud.

Basically, the aim of JFrog is to give you one
solution to go from development, which we have done very well up
until now – all the way through the cloud, and to distribution –
using fully integrated tools, of course.

Author
Comments
comments powered by Disqus