Hacker News Reader

Comment by eriknstr | Hacker News Reader

OP is not suggesting ZFS above all else.

> And every enterprise system has already moved way past what ZFS can do, including enterprise-class offerings based on ZFS from Sun, Nexenta, and iXsystems.

However, OP is naturally suggesting ZFS over NTFS, HFS+, ext3/4 and even ReFS and APFS.

> Still, ZFS is way better than legacy storage SOHO filesystems. The lack of integrity checking, redundancy, and error recovery makes NTFS (Windows), HFS+ (macOS), and ext3/4 (Linux) wholly inappropriate for use as a long-term storage platform. And even ReFS and APFS, lacking data integrity checking, aren’t appropriate where data loss cannot be tolerated.

View on HN

ReFS has data integrity checking and repair.

...but ReFS File Integrity is disabled by default! Integrity streams are enabled for metadata only. Here's how to turn it on: https://technet.microsoft.com/en-us/library/jj218351%28v=wps...

Interesting. I'd like to know what the "enterprise" has moved on to from ZFS.

Enterprise Storage == Real Storage Arrays

Basically every storage array is years beyond ZFS in terms of features and capabilities. Even the ones based on ZFS! And I don't think anyone would argue that. After all, Dell EMC, HPE, Pure Storage, HDS, etc spend many millions a year developing advanced storage technologies.

I'm finding this difficult to believe; any references for some of the features and capabilities "years beyond ZFS" that are in a shipping, supported, enterprise-level commercial product?

http://www.scality.com/products/ring/

I realize that not everyone lives in the world of enterprise storage, so I appreciate your question.

ZFS is a filesystem (and also kind of a volume manager) and is designed to be fairly versatile and to run alongside an OS and applications. ZFS includes some impressive features long found in enterprise arrays, including snapshots, mirroring, replication, RAID, and data protection.

Dedicated storage arrays have long supported many more specialized features, and I urge you to read up on cool products like Pure FlashArray, NetApp SolidFire, EMC Unity, EMC Isilon, HPE Nimble Storage, Tintri, etc. You'll see that most/all have awesome hybrid (flash/disk) capabilities, all-flash tuning, integration with VMware, Windows Server, and Docker, scale-out capability, etc.

It's a pretty amazing field, and that's why I've dedicated my professional career specializing in enterprise storage. Maybe a good place to start is the Storage Field Day video series on YouTube: https://www.youtube.com/results?search_query=%22Storage%20Fi...

There are all-flash based and hybrid ZFS configurations from some vendors, so I don't understand how that's a point.

As for the rest, I was looking for more concrete comparisons, and third-party software integrations don't seem relevant to this conversation.

> ZFS includes some impressive features long found in enterprise arrays, including snapshots, mirroring, replication, RAID, and data protection.

This is why NetApp sued Sun in 2007, claiming patent infringement.

I worked in a data center that stored medical records around that time. We were moving off of a StorageTek PowderHorn 9310 (that thing was mesmerizing) to some multi-cabinet EMC spinning disk array. You couldn't get access to that thing without hooking up a FC HBA to the SAN. Fast-forward a few years and I found myself in a similar job at a much smaller e-commerce firm. We had some old finicky NetApp filers we ended up getting rid of in favor of Sun Storage 7000 storage appliances running ZFS. Both of them supported replication, but the Sun Storage appliances were far better.

Yeah, SAN and NAS are two different things, but there's a lot of money, a lot of shared technology, and a lot of similar use cases that have driven convergence over the last decade or two.

http://www.pixitmedia.com/pixstor/

basically lots of nice python wrapper around GPFS.

ZFS isn't clustered and has no concept of HSM, so migrating hot blocks from NVMe to SSD then disk and tape isn't an option

Also stuff around disk rebuilding is still pretty old. Modern FSs can detect a failed disk and rebuild the array in minutes based on FEC and even distribution of data.

The documentation for GPFS is spectacular as well. IBM's redbooks are a beacon of clearness in a world of shart.

ZFS isn't clustered

The ZFS storage appliance offers clustered configurations: https://www.oracle.com/storage/nas/index.html

It's unclear what your requirements are when stating "clustered".

and has no concept of HSM, so migrating hot blocks from NVMe to SSD then disk and tape isn't an option

Oracle HSM integrates with ZFS:

https://www.oracle.com/storage/tape-storage/hierarchical-sto...

Also stuff around disk rebuilding is still pretty old. Modern FSs can detect a failed disk and rebuild the array in minutes based on FEC and even distribution of data.

ZFS can detect a failed disk and rebuild in minutes as well depending on storage and system speed, etc. so I'm wondering what you're referring to?

Reply on news.ycombinator.com