Castor: CERN Advanced STORage Manager

Posted by naves 5 days ago

Comments

Comment by tempay 5 days ago

I’m a little confused by this submission. CASTOR is the old system that has since been replaced by the CERN Tape Array since ~2020: https://cta.web.cern.ch/cta/

This is mentioned on the page but it’s easy to miss.

For the current status of tape storage at CERN see: https://indico.cern.ch/event/1471803/contributions/6967379/a...

For reference, most disk storage for physics data uses an in-house solution called EOS: https://eos-web.web.cern.ch/eos-web/

Comment by bfkwlfkjf 5 days ago

Does tape array replace castor? Just from the names it sounds like tape array is the actual storage, and castor is an abstraction that automatically decides what's kept on disk and what's kept on tape

Comment by tempay 5 days ago

The abstraction isn’t really a thing any more. It was a nice idea but in practice it’s an operational nightmare not knowing if data is available and for how long it will be. For reference staging can take days during intense activity and you don’t want to loose performance randomly seeking around and switching between tapes.

Comment by sam_bristow 5 days ago

The linked page seems to think it does.

"As of June 29th 2020, CTA, the CERN Tape Archive, started to be operated as the successor of CASTOR and gradually replaced it."

Comment by john_strinlai 5 days ago

looks like the image on the right is broken, but it is supposed to be: https://cta.web.cern.ch/cta/assets/images/namespace_statisti...

(looks like this submission uses https://castor.web.cern.ch/content/home.html instead of https://castor.web.cern.ch/castor/ the second link does not have the broken image)

Comment by pezezin 4 days ago

Fun fact: CERN sells old data tapes as souvenirs, I got myself one of the old LHC tapes :)

Comment by zerr 4 days ago

With the data included (not wiped)?

Comment by pezezin 4 days ago

According to the label, the data is included. But it is just experimental data from the LHC, nothing that has to stay secret.

Comment by tempay 4 days ago

As far as I know nobody has ever tried to read one ;)

Comment by tempay 4 days ago

Though 5PB of data is freely available via https://opendata.cern.ch/

With another 4PB available which can be processed on request to extract samples of interest: https://opendata.cern.ch/docs/lhcb-releases-service-to-acces...

Comment by _pferreir_ 4 days ago

As others have said, CASTOR has been discontinued, and replaced with CTA:

https://gitlab.cern.ch/cta/CTA

Its memory is still alive in CTA, however:

https://gitlab.cern.ch/cta/CTA/-/blob/main/catalogue/TapeSea...

Comment by Davidbrcz 4 days ago

I was an intern at CERN in mid 2010s and worked on this !

Comment by dokyun 5 days ago

Wonder how this compares to Venti[1]. It looks a lot more complicated (not really a good thing).

[1]: https://doc.cat-v.org/plan_9/4th_edition/papers/venti/

Comment by linksnapzz 5 days ago

You could use tape as a backing for Venti arenas; don't know if anyone ever did so. The original Bell Labs fileserver used an MO jukebox for WORM archives, which today LTFS tape is a pretty close approximation of.

Comment by Melatonic 5 days ago

This is actually super useful for real world stuff. Thanks for this.

Tape is boring but when an intern / AI / tectonic plate accidently destroys your database setup it is a huge lifesaver

Anybody know what these fancy Oracle tapes are? Is it just their implementation of a regular standard?

Comment by tempay 5 days ago

See this conference talk from last week: https://indico.cern.ch/event/1471803/contributions/6967379/

There isn’t a recording but slides at linked from that page.

Comment by linksnapzz 5 days ago

If it's Oracle Tape, it's proprietary T10000-series 1/2in linear tape and associated drives, that they got when they absorbed Sun (and Sun got when they bought StorageTek). Multiple vendors made tape media for these, but they were not compatible w/ LTO tape nor the IBM 3590-series enterprise tape format.

Comment by perlgeek 5 days ago

"Castor" was the name of a storage system used for transporting nuclear waste in Germany. There were quite a few protests against shipping nuclear waste through the country.

Wouldn't have been my choice for a software project :-)

Comment by tempay 5 days ago

It’s also French for Beaver which is more likely the origin of the name.

Comment by rzzzt 5 days ago

It's also Latin and Greek for beaver which is more likely the origin of the name.

Comment by tempay 5 days ago

Latin and Greek aren’t one of the working languages at CERN (French and English are)

Comment by randiantech 5 days ago

also spanish

Comment by elashri 5 days ago

I would say "Italian" :)

Comment by pjmlp 4 days ago

And Portuguese. :)

And to keep this thread, I think our three languages should count as one, because at least 20 years ago, it was quite common to have Portuguese, Italian and Spanish mingle in several activities.

Source, ATLAS TDAQ/HLT Alumni.

Comment by elashri 4 days ago

I was just commenting based on the cliche that Italians are everywhere at CERN. So you will always hear Italian language.

Comment by pjmlp 4 days ago

They are, that is how learnt Italian without much effort. :)

Comment by antonvs 4 days ago

Castor oil makes you poop, maybe there’s a data management metaphor in there somewhere.

Comment by linksnapzz 4 days ago

I imagine that an ancient roman would think "Oleum Castorum? It's either oil you get from rendering beavers, or oil you use to lube your beaver..."

Comment by adev_ 4 days ago

A few historical additions for anybody interest:

- CASTOR at CERN had also its disk centric derivative named DPM (Disk pool manager) that helped to power the LHC computing grid for multiple decades (WLCG) before getting deprecated.

- Interestingly DPM had an architecture quite aligned with the original Google File system even if developed completely separately: (One metadata node, multiple disk node. Design to do Write-once-read-many with very partial POSIX semantics).

- The LHC computing Grid is an association of research centers with their own infrastructure. As such, they had (historically) many diffent storage systems with diffent protocols and interface.

- To unify this madness, an attempt to do a "standard" protocol was made in the 2000s: the SRM protocol (storage resources manager). In a pure XKCD fashion, it went as bad as you can imagine. It tried to rely on the tech of the time (XML, SOAP, WSDL) and is a school case of terrible protocol design (bloated, slow, weak consistency, massive server overhead, stupidly complex to implement and quite insecure). The spec are worth a read if you want a good laugh [1].

- After 20y of struggle, SRM was eventually dropped for a more pragmatic and ad hoc solution based on HTTP + xrootd [2]. EOS itself uses xrootd quite extensively. (if this did not change)

- The history of computing at CERN is globally interesting because it is a pretty good image of the evolution of computing and of the "tech fashions" associated with it.

[1]: https://sdm.lbl.gov/srm-wg/doc/SRM.spec.v2.1.1.html

[2]: https://xrootd.org/

Comment by boznz 5 days ago

The various CERN web pages such as this were a treasure trove of information when I was working on my last novel. I actually included a few paragraphs on Castor thinking of using it as a side-plot, but my editor cut the plot out along with a few other technical niceties. Sigh!