Spectrum Scale 5.0.3 has been released, and comes with a number of enhancements. In this article, I will highlight a few of the thing that you can look forward to after upgrading to 5.0.3.

To start with, there is the new Clustered Watch feature, which enhances and builds upon file audit logging and watch folder functionality by allowing you to:

  • Watch file operations across clusters using a single scalable, resilient, centralized tool
  • Similar to file audit logging, clustered watch monitors activity on a file system.  Additionally, clustered watch can monitor activity on a fileset or inode space
  • Events are consumed by a conduit that is configured to send these events to an external ‘sink’
  • The external ‘sink’ is an externally managed Kafka queue outside of the Spectrum Scale cluster 
  • Allows greater flexibility to handle and react to file events

Further enhancements were made across the entire Spectrum Scale product. A few of those highlights are:

Filesystem

  • CCR Recovery Enhancements – additional options to help with broken cluster recovery
  • Support for additional file compression libraries for genomics workloads
    • Support for the new compression libraries: zfast, alphae, and alphah
  • Enhancements to mmnetverify for RDMA networks
  • Linux filesystem stat enhancements for compatibility
    • Report in units of blocks or subblocks for filesystem sizes and free space
  • Enhancement to mmkeyserv to show certificate expiration dates
  • QoS improvements in large clusters – data collection frequency is dynamically changed based on cluster size
  • Performance improvements for file creation, specifically for a large number of nodes simultaneously creating files in the same directory
  • NSD servers periodically poll internal disks for I/O errors
  • Fsck improvements
    • Estimate run time without scanning inodes
  • Repair block allocation map without taking a filesystem offline

GUI

  • Quota management enhancements
    • Define capacity and inode quotas for filesets, users, and groups
    • Enhanced default quota management to define and manage default quotas for filesets, users, and groups
    • Improved usability to perform quota tasks such as enabling and disabling quotas, setting quota scope, grace time, etc
    • Introduced daily quota reports, where recipients receive reports of exceeded quotas
  • GUI support for Integrated Archive Mode (IAM), to support file retention policies
  • New panels to configure the GUI to utilize an external LDAP server for GUI user access management
  • Navigation and menu improvements
  • Default user is no longer enabled (security enhancement)
  • Improved management and monitoring of NFS and SMB shares
  • New GUI wizard for configuring and modifying file (SMB/NFS) authentication
  • ILM policy editor has enhanced support for external pools

RAS and System Health

  • Proactive Network Reconnect for improved network resiliency
    • Attempts to reconnect to a remote client if a connection is slow or stalled
    • See mmchconfig man page for details on how to enable this feature
  • Callhome improvements
    • Simplified setup
    • Improved scalability and performance
    • Heartbeat data collection
    • Trigger callhome on certain critical events
  • NVMe drive monitoring
  • mmhealth usability enhancements
  • mmhealth state resync to force synchronization of status across the cluster
  • Filesystem maintenance mode improvements
  • NFS enhanced health checks for lock manager
  • CES protocol services are no longer automatically started on suspended nodes

REST API

  • Filesystems can be mounted or unmounted using the API
  • Ability to view and update default quotas on filesets

SMB and NFS protocols

  • GUI support for NFSv4 pseudo path
  • Support for SMB 3.1.1
  • Improved handling for the SYNCHRONIZE ACL bit over SMB

Transparent Cloud Tiering / Cloud Data Sharing

  • Automatic container spillover feature
    • Use another cloud container when enough objects have been migrated to the first
  • Microsoft Azure support
  • Client assisted recalls for greater speed and flexibility when recalling files

Deployment Toolkit and Upgrades

  • SLES 15 support for install, deployment, and upgrade for x86 and Linux on Z
  • Upgrade flow changes to minimize I/O disruptions
  • Mixed OS support
  • Pre-checks determine exactly what packages must be upgraded, and post-checks ensure that all packages have been upgraded
  • Enhanced dependency checking

AFM

  • New configuration parameters to configure refresh of cache data in asynchronous mode

Big Data & Analytics

  • Support for Hortonworks Data Platform 3.1 and Management Pack 2.7.0.2
  • Support for SLES 12 SP3 for new installs on x86_64

Spectrum Scale on IBM Z

  • Clustered Watch support
  • Transparent Cloud Tiering support
  • IBM Storage Enabler for Containers
  • KVM host support
    • Spectrum Scale can now be utilized for KVM storage pools (directory pool only) for virtual disks (qcow2 and raw) on a KVM on IBM Z host
  • Currency:
    • SLES 15 support

Join The Discussion

Your email address will not be published. Required fields are marked *