Attala Scale-out Automated Orchestration Software

Any good scale-out infrastructure is only as good as how easy it is to manage and orchestrate on behalf of the multiple tenants using the infrastructure.  At Attala, we have created SPARA which is our product acronym for Security, Performance and Resource Automation.  It addresses all aspects of the management lifecycle: (i) infrastructure scalability via automated discovery, (ii) zero-touch allocation of resources to users (no human intervention required), (iii) performance monitoring and diagnostics and (iv) scale-out infrastructure health reporting, alerts and diagnostics.  It is technically challenging to automate an infrastructure that is distributed and continues to scale-out.  And that's exactly what the Attala SPARA software is designed for.



Zero-touch multi-tenant provisioning

In a multi-tenant environment, each tenant has their own GUI or API-driven interface to provision NVMe resources for their applications.  Admins or tenants simply request the namespace size and class-of-service for their application needs.  The SPARA software autonomously selects the resource that satisfies the request while also optimizing infrastructure utilization.  With additional plug-ins, Cinder, Kubernetes and other tenant interfaces will also leverage SPARA’s automated provisioning


Multi-tenant Performance Analytics & Automation

The SPARA solution tracks the performance of thousands of applications in real-time to ensure they get the resources and performance they need.  The latency, IOPS and throughput of application traffic to every NVMe-oF volume in the cluster are instantly available for visualization, analysis and comparison to the volume’s existing QoS settings to ensure the application is getting what it needs or to reclaim unused resources.  In addition, live performance data can be stored for historical analysis or alerts generated for immediate response.

Orchestrate Persistent, Shared NVMe Volumes

Beyond provisioning of raw NVMe volumes, SPARA also manages persistent volumes – that contain data that persists between being used by one or more hosts/initiators.  From a library of persistent volumes, tenants are able to select and attach these data-laden NVMe volumes – either to a specified initiator or to several initiators using NVMe volume-sharing.  This capability enables instant rebuilds of failed scale-out database instances.  And also eliminates lengthy copies and redundant costs via data-sharing between multiple initiators that need access to the same data.

Infrastructure Health Monitoring and Troubleshooting

Get an at-a-glance summary of your entire infrastructure health via an intuitive dashboard, designed to let you drill-down for more information with just a few simple clicks.  Using this single dashboard, you can view metrics for the physical infrastructure and tenant statistics.  Isolate problems and troubleshoot quickly with simple workflows, granular detail and extensive search capabilities.

  • Check alerts by data node, network ports, NVMe SSD (SMART), or any other attribute – and then click the device for more details and diagnostics

  • Get information about connected hosts, data nodes for fast troubleshooting.

  • Manage either via the SPARA GUI or via RESTful APIs

Automated infrastructure Discovery & Configuration

The solution autonomously discovers newly added host-NVMe-oF adapters or storage nodes using the SSDP protocol and periodic polling.  Storage node discovery is autonomous with detection of installed NVMe SSDs and network ports. Configuration is autonomous with assignments of IP and subnet addresses, unique NQN names, etc.  Minimal manual intervention is required before a new device is ready for use.

Alerts & Reporting

Store your management data indefinitely so you can create reports containing historical data if needed.  Schedule reports and receive automated email report

Maintenance and Firmware Management

SPARA handles firmware updates across all host NVMe-oF adapters and storage nodes. The new firmware will apply after a user-specified reboot.

Secure and Always-on

The Attala infrastructure is protected from malicious intent via HTTPS protection on all management traffic and authentication is via login, password and per-session API-keys.  
Tenant and operators are assigned unique login and credentials.  For each tenant, the operator also defines role-based permissions.  LDAP authentication and active-active SPARA software instances are also supported.

Managed End-points

  • All Attala Host NVMe-oF Adapters (HNAs)

  • All Attala Storage Nodes

  • Mellanox ConnectX4 RNICs or ordinary L2 NICs

Orchestration Framework - Northbound

  • Attala GUI and REST APIs

  • OpenStack Cinder (planned)

  • Kubernetes (planned)

Fabric Management Standards – Southbound

  • DMTF Redfish 1.0, SNIA Swordfish 1.0

  • MQTT v3 (Monitoring), IETF SSDP (Discovery)

Infrastructure Management

  • Visualization of all HNAs and storage nodes - health and resources

  • NVMe SMART statistic drill-down & alerts

  • Asynchronous, MQTT-based alerts for uncorrectable errors (Ethernet link failures, HNA/NSC temp) – displayed on GUI

  • Consolidated system & error logs

  • Storage node chassis monitoring: Temperature

Storage node chassis management

  • LED beaconing

Performance Monitoring Statistics

  • Read-latency (min, max, average), Write-latency (min, max, average)

  • Read-throughput (average), Write-throughput (average)

  • Read-IOPS (average), Write-IOPS (average)

SPARA System Requirements

  • Bare-metal or Virtual Machine

  • Network access to management network

  • Red Hat or Suse Enterprise Linux