A napp-it ZFS Storageserver includes many features of expensive High-End storagesystems like realtime checksums on data and metadata with online scrubbing, CopyOnWrite with snaps and sophisticated readcaches in RAM and on SSDs (Hybridstorage).
Features of napp-it (overview): Featuresheet.pdf
Service relevant features
- Solaris and free derivates like OmniOS are specialized and developed around ZFS for bootsystem and datapools.
with the commercial use for many years this results in a very stable
and sophisticated ZFS storage system. New ZFS features are done beside
Oracle from Illumos and included in OmniOS quite fast.
- Filebased sharing (AFP, FTP, NFS, SMB or WWW) as a filesystem property
- Kernelbased SMB high performance server with Windows Share/File
level ACL, Snaps as Windows previous version and AD integration. Unique
support for Windows ACL with Windows SMB groups and AD or Server based
security id's (SID) as ZFS file attributes. (SAMBA is optional but does not support Windows SID as a file-attribute)
- Blockbased shares (MPIO, iSCSI, FC) either with the universal Comstar framework or as filesystem zvol property.
- Network virtualisation with Crossbow for virtual switches and virtual network adapters on vlans.
- Blockbased highspeed filesystem replication locally or over the
network.with zfs send/ netcat in wirespeed with all relevant attributes
like ACL, Windows SIDs or shares. After the initial transfer only
modified datablocks are transferred based on snapshots. This allows near
realtime replication of filesystems over the network.
Filesystem relevant features
- No relevant limits regarding capacity, number of disks or filesystems or files or folders.
- ZFS is designed for best of all data security and performance with hundreds od disks and Petabyte storage.
- Always consistent filesystems due CopyOnWrite (even on a crash during a write, no offline chkdsk needed)
- Data versioning with read-only snapshots. Nearly unlimited snapshots can be created without significant delay or initial space consumption due CopyOnWrite allowing revision safe storages.
- Always verified data with end to end checksums on OS Level on data and metadata. With redundancy all errors are repaired during read or on regular online scrubs. This allows safe long term storages with ZFS.
- Poolbased Storage Virtualisation with dynamically sized filesystems. Management is done with reservations and quotas. Volumemanagement is part of ZFS.
- Transparent compression without performance degrations due the LZ4 compressor and realtime deduplication.
- Windows compatible NFS4 ACL with inheritance. Case (in)sensitive or mixed filesystems selectable.
- Secure sync write on filesystem level selectable to avoid dataloss on powerloss. Optionally you can add a highspeed dram (ZeusRAM) or SSD based logdevice (ZIL) to ensure secure writes without a performance degration.
- Small random writes are collected in a RAM write cache and transformed to large and fast sequential writes. This improves write and general I/O performace.
- In addition to the ARC Read Cache that caches data in free RAM you can use SSDs as L2ARC cache. The sophisticated ARC/L2ARC is highly optimized and works on datablocks. (Hybrid Storage to accelerate disk based storage).
- Storage from several vdevs (Raid-arrays) is managed as a datapool. You can extend the pool by adding vdevs/ raid-arrays.
- Controller independent Software Raid. You can import a pool on any supported controller even on other Operating systems. You do not need special hardware raid controllers with CPU, Cache or BBU. Best are simple HBA controllers in raidless IT-mode.
- No write hole problems like with Raid 5/6 where a powerloss can lead to a corrupted filesystem. Supported Raid levels are Raid 0, 1, n-way-mirror, Z1, Z2, Z3. ZFS reads from all mirrors simultaniously and stripes data over vdevs/ raid-arrays like a multi-raid-60. With Z1 a single disk failure is allowed similar to Raid-5. Z3 allows a 3 disk failure per vdev without a dataloss.
- No problem with multiple short disconnects of several disks in a raid due to cable or power problems. This leads only to a short degraded or offline state. If enough disks come back the pool goes online without the need of a raid-rebuild or offline chkdsk.
- Hot-Spare disks. Failed disks are replaced automatically with hot-spares.
- Raid-Management is integtrated in ZFS
- Sequential performance on ZFS scales very well with total number of data-disks in a pool.
- I/O performance of ZFS scales very good with number of vdevs in a pool. With SSD only pools you can easily reach thousands of I/O per second with very low latencies. This allows high performance pools for database or ESXi use.
- All from one hand. Beside OSX and Windows, Solaris and its free derivates are the only operating systems "from one hand". All what is needed for a storage server with all services like NFS, SMB or iSCSI is maintained from Oracle for Solaris or Illumos for the free Solaris derivates and not different/maintained on distributions. Dies gives you a very homogenous handling compared to other platforms like Linux.
- Easy setup within minutes (one command after OS installation) with Web-UI for the whole storage management with job-management for own jobs, scrbbing, snaps, replication or status/alert emails.
- nonfree extensions for ACL management, Monitoring or Remote Replication.
- Napp-it supports Linux basically (Debian, Ubuntu LTS) with ZFS management and remote replication.
- Active Active HA is optionally available from high-availability.com