Unveiling the Evolution: Proprietary Hardware to Agile Software-Defined Solutions

Data Center (SDDC)
© 2023 StoneFly, Inc. | All rights reserved.
Software-Defined
Explore software-defined data center technology and
its integration into StoneFly enterprise hyperconverged
solutions.

StoneFly USS at the Core
Core Review of a Software Deﬁned Data Center (SDDC) Appliance
Today’s data centers and Cloud have moved beyond the rigid deployment of
proprietary hardware based compute, network and storage solutions. To maximize
hardware utilization, reduce complexity and management and environmental cost
well-proven software virtualization solutions for compute - as provided by hypervisors
- to full cloud automation software platforms. A complete software-based infrastruc-
ture solution requires more than compute virtualization, however. Combining the
agility of cloud and virtualization software, along with Software- Defined Networking
(SDN), and Software-Defined Storage (SDS) solutions are becoming the cornerstones
of the modern data center. We see these elements coming together in software to
enable the greatest data center agility, by allowing the software to shape the under-
lying hardware to deliver services in the best form for applications to consume. By
decoupling software from the underlying platform, we also enable the greatest
choice in platform flexibility. This provides a quantum step in reducing the overall
cost (CapEx & OpEx) as embedded Hypervisor allows multiple Virtual Machines (VMs)
to run on a single physical host or multiple cluster hosts. They mediate all I/O opera-
tions including read and write requests to centralized or network attached NAS and
SAN storage arrays. These arrays are typically used to provide shared storage for all of
the VMs.
StoneFly USS (Unified Storage & Server) is a hybrid hyper-converged appliance
that takes a simpler approach to converged architecture. It does this by incorporat-
ing local direct-attached storage for direct, faster performance and greater flexibility
while incorporating block (SAN) and file (NAS) storage with extensive data services
with the advance software-defined network as interconnect communication and
routing. Each node in a StoneFly can be standalone or part of cluster(s) of scale out
nodes. Each node includes embedded hypervisor Which integrates with customer’s
virtual environment such Hyper-V (System Center) or VMware ESXI( Vcenter), Raid
with associated enterprise SAS 7200,15K grade hard disk drives for low-cost high-ca-
pacity storage disk and optional flash-based storage to deliver massive IOPS for
high-performance.
1

Cloud Connect One™ (Microsoft Azure, VMware vCloud, and so forth) can be used for
extending the storage, elasticity, provide higher business continuity, lower CAPEX
(Capital Expenses) and OPEX (Operating Expenses), and adhere to the principles of
software-defined provisioning of resources.
StoneFly Implements all Control Logic as a Software Service
Virtual storage controller(s) run on each cluster node greatly improving scalability
and resilience. Since the storage and control logic is now local to the guest VMs,
performance enhanced and bottlenecks eliminated.
Expensive centralized storage or dedicated storage networks are no longer needed
since the Distributed File System aggregates local storage across all of the nodes.
There is no metadata database to limit the quantity of StoneFly USS nodes partici-
pating in the cluster allowing scale out. StoneFly USS appliances use a hash-based
Elastic Hash Algorithm in each node to accomplish this task. As a result, StoneFly
USS appliances do not experience the latency issues that associated with competing
cluster solutions that utilize metadata databases.
StoneFly storage controllers can provision as a block, file, and object. StoneFly can
present a file as a file, or a file as an object per the client’s request. StoneFly’s intuitive
NAS software allows users to create a single storage pool that can partition into one
large volume with a single global namespace, or several smaller volumes for applica-
tion or tenant partitioning. These Datastores presented to the hypervisor. These
volumes are also fully POSIX-compliant which allows for seamless integration with
any applications requiring this capability.
Using the standard NFS, CIFS/SMB, iSCSI, Fibre Channel and InfiniBand protocols to
provide storage for all hosted VMs, the hypervisor communicates with StoneFly’s
software exactly as it would be a traditional storage array, and no changes need are
made to the virtual environment. VMs are provisioned and managed exactly as
before, but without having to configure LUNS, volumes, and RAID groups if desired.
2

The StoneFly NAS or SAN storage can optionally be provisioned to internally-hosted
VMs on each appliance node, provisioned to externally-hosted VMs running on
VMware, Hyper-V, Xen, or Oracle VM platforms, as well as provisioned to external
physical hosts via the LAN or WAN.
When a guest VM needs to write data as an I/O transaction, its request passed to
StoneFly’s virtual controller which then executes the write to local flash storage. It is
critical that the data are fully protected so that it is always available even if a node
fails. The strongest levels of data protection ensured by having typical read operation
request replicated synchronously across multiple nodes. These requests served by
local storage resources where the data has cached to the fastest user-selected
resource such as flash. Keeping data local to the VM or Hosts provides the fastest
possible performance because for the majority of read requests the data never has to
traverse the network.
StoneFly supports the Automatic Storage Tiering. The “cold data” (less frequently
requested data) migrates from flash memory to lower cost hard disk.The Data is
migrated back to flash if it once again becomes “hot data” (requested more frequent-
ly data) by a VM or host.
StoneFly supports standard high availability capabilities such as VMware vSphere
High Availability and Fault Tolerance, or Microsoft Hyper-V clustering. In the event of
a node failure, a VM will auto-start on another node. If that VM or Host needs to read
or write data, then it sends its request to the local StoneFly controller. If the data
access requested is not locally available, then the controller determines where replica
data has stored and forwards the request to the correct StoneFly controller node(s).
The data then sent back to the local StoneFly controller over a standard Ethernet
network whereby the local StoneFly controller will pass the data to the VM through
the Hypervisor for future local storage access.
The StoneFly software once again replicates the data throughout the cluster to return
the full cluster system to a fault tolerant state. Each StoneFly node runs inde-
pendently and leverages the StoneFly distributed software architecture to create a
completely different unified cluster as a virtual computing platform.
3

StoneFly USS appliances can provide a Scale-Out Storage platform that easily man-
ages unstructured data for physical, virtual, and cloud environments. Ability to work
with File, Block, and Object storage within a scale-out architecture, StoneFly appli-
ances can address different type of workload and manage PetaByte-scale data
growth in a highly cost-effective manner. StoneFly delivers an extendable fabric
across physical, virtual, and cloud resources so customers can transform their big,
semi-structured and unstructured data from a burden into an asset.
StoneFly’s USS appliance stores various types of unstructured data including:
• Rich media content like videos, images, and audio files
• Backup images and Nearline storage archives
• Big data — log files, RFID data, and other machine-generated data
• Virtual Machine images
StoneFly eliminates storage silos by enabling global access to data through multiple
file and object protocols. StoneFly’s USS appliances can deploy in a single node or
multi-node cluster configurations. Each node’s storage can be “Scaled Up” with one
to ten expansion units in increments of 2U (12/3.5” drives or 24/2.5” drives), 3U (16/3.5”
drives) and 4U (24/3.5” drives) per chassis and up to 8TB per drive. Cluster nodes and
their respective expansion units can be “Scaled Out” by seamlessly adding new nodes
with zero downtime while maintaining or perhaps even increasing performance
levels. With StoneFly USS, businesses can start small and easily expand to achieve
truly large scale storage.
“Scale Up” and “Scale Out” NAS: File and Object Storage
StoneFly USS appliances deploy global scale out storage for OpenStack Block or
Volume Storage (Cinder), OpenStack Container or Object Storage (Swift), as well as
storage for the OpenStack Image service (Glance).
Global Scale-Out Storage for OpenStack – Cinder, Swift, Glance
4

Customers can mix various models of StoneFly USS appliances within a single cluster
with each node supplying a different ratio of storage and compute resources. Cus-
tomers scale compute and storage within the cluster to support their specific work-
load.
Heterogeneous Clusters: Mix and Match StoneFly Appliances
StoneFly’s unique policy-driven caching capabilities ensure the appropriate level of
accessibility and performance based on the data stored. Featuring the choice of
write-through, write-around and write-back caching modes with or without system
buffer cache, StoneFly’s Flash Cache is fully flexible to allow users to customize their
systems to reach that perfect balance of performance and redundancy.
StoneFly USS appliances with Flash Cache can be used as on-premises cloud acceler-
ation gateways. The cache sits locally on the appliance, delivering the fast perfor-
mance users expect while asynchronously replicating to the cloud.
Select what percentage of the caching device you wish to use for storing metadata
and cached blocks.
Using the cloud for primary or secondary storage? Configure a cache so that 100
percent of your data sits locally as well as in the cloud.
StoneFly Flash Cache
Data is fingerprinted using an SHA-1 hash on ingest. This fingerprint information
stored in StoneFly’s metadata layer. Deduplication operations are software-driven
but leverage the hardware-assist capabilities of the Intel chipset for an SHA-1 hash
generation, allowing for the fastest possible performance. This approach is ideal for
Elastic Deduplication: Fine-Grained Parallel Deduplication
5

nearly any virtual workload, including virtual desktops and VMs, computing deploy-
ments, big data applications, media recording and broadcasting, DR initiatives, and
more.
Redundancy at its Greatest: Dual Cluster Nodes and Dual Active-
Active Hardware RAID Controllers
The USS-HA series of StoneFly’s hyper-converged product line supplies a highly-avail-
able cluster solution. Each unit is 5U Rackmount; each USS-HA consists of two
Hyper-Converged 1U Controller clustered, failover engine with an embedded hypervi-
sor and integrated StoneFly Virtual Storage Appliance engine (Tops units). The third
chassis is 2U (12Bay) or 3U (16 Bay) RAID subsystems with dual active-active failover
hardware controllers.
The clustered head nodes offload and process storage protocol services such as iSCSI,
NAS, Fibre Channel, and data services such as snapshot, thin provisioning, campus
mirroring, encryption, deduplication, replication, Cloud Connect, as well as Hosting
Virtual Machine using the Embedded Hypervisor (ESXi or HyperV) The active-active
hardware RAID controllers manage the entire RAID, allowing rebuilds, expansion,
and failovers at the highest performance.5U version of the StoneFly USS-HA with 2
cluster nodes and 16-bay storage array
A detailed diagram of the two-node cluster system with shared storage protected
with dual failover RAID controllers.
6

In a world increasingly driven toward a geographically dispersed and distributed
environment, more and more organizations have turned to the cloud to take advan-
tage of its globally accessible pay-as-you-go business model with reduced mainte-
nance costs. The allure of cloud storage is even stronger. Resource-strapped IT orga-
nizations are tired of over-purchasing storage today and finding themselves to be
under-provisioned in just a couple of years. They’re devoting a hefty amount of time
and money to storage upgrades, decommissions, migrations, tape backups,
remote site visits – only to start the cycle over again, and again … and again. Not to
mention the sleepless nights worrying about whether their disaster recovery strategy
will hold up if it is ever needed.
Despite the obvious benefits of cloud storage, many organizations remain wary of
moving to the cloud. Traditional storage may be costly, resource-intensive and cum-
bersome, but it also performs well, is secure, highly available, and just works.
Cloud Connect One™: One-Step Connection to the Major Public
and Private Clouds
That is where StoneFly’s cloud-integrated storage solutions come into play. Designed
to combine the resource efficiency of the cloud with the peace of mind you get from
traditional on-premises storage, StoneFly’s USS Appliance product line delivers enter-
prise-class storage without the enterprise-sized headaches.
Extend Storage Capacity and Regain your Freedom.
StoneFly USS appliances employ both in-flight and at-rest encryption to ensure the
security of your data. Each volume can be configured using AES 256-bit encryption
before replicating your data to the cloud. Also, encryption keys are stored locally with
the customer, thereby preventing unauthorized access by either a hacker or the
cloud storage provider itself.
Security and Encryption
7

One of the best utilizations of cloud storage is for disaster recovery.Traditional disaster
recovery strategies involve either secondary sites or backups to tape or disk. For
many businesses, a secondary site can be prohibitively expensive. Backups to tape or
disk are highly manual, slow, and can be unreliable.
With StoneFly’s USS hyper-converged appliances, organizations that experience a
disaster can avoid downtime by quickly deploying a secondary StoneFly appliance
running at a remote site or running in the cloud to access their data. Since StoneFly
appliances are available virtually, in cloud compute environments, and physically,
organizations can recover and access their data within a few minutes.
Local systems can mount remote iSCSI drives or NAS shares via the StoneFly Cloud
Drive in Microsoft Azure.
One Click Disaster Recovery
8

Cloud Connect One™ Scale-Out Cluster.
Remote Replication from an on-premise StoneFly Appliance
to Microsoft Azure.
9

Backup and disaster recovery option.
Enterprise-class features available with StoneFly USS hyper-converged appliances:
• Full Block - Interchangeable File and Object: Store as one type and access with
the other type.
• 200 iSCSI, 512 Fibre Channel Volumes Supported with 1022 Concurrent Host iSCSI
Sessions per Node
•Port Teaming, Failover, and Load-Balancing
StoneFly USS Hyper-Converged Appliances: Available Features
10

• Volume-Level Access Control and Dynamic Volume Management
• Support for iSCSI, Fibre Channel, SNMP Traps, UPS, Nagios Monitoring, Reporting,
RAID Monitoring, Call Home
• System-Wide Graphical Performance and Utilization Reporting
Available Features – Continued
• Advanced Report Management
• Automated Online Volume / Storage Expansion
• StoneFly Snapshot Services with 2520 Delta-Based Snapshots per Subsystem
• Mountable Read-Write Snapshot Volumes
• Snapshot Schedule Utility, Command Line Interface Utility
• StoneFly Real-Time Synchronous Mirroring of iSCSI / Fibre Channel Volumes and
Nodes (Campus Mirroring)
• One-to-Many, Many-to-One, Many-to-Many Asynchronous Replication
• Multi-Site/Multi-Appliance Replication and Unified Central Management System
• Thin Provisioning with Space Reclamation
• Tiered Storage Architecture with Hardware and Software Support
• Hardware-Enabled Block Level AES 256 Volume Encryption
• Fibre Channel SAN Target (License Bundled with choice of 4Gb/8Gb/16Gb HBAs)
• Hardware-Assisted and Optimized Block Level Parallel Data Deduplication with
High-Performance Index Engine
• Support for SATA, SAS, SSD
• Tiered Storage Support at the Block Controller Level
• Flash Cache Support for Fully-Tiered Storage: SATA-> SAS -> SSD -> CLOUD
• Flash Cache using Flash SSD for High-Performance IOPS Applications
• Cache Support of up to 1TB per Node
• Up to 36 Cores Storage Processor and Total of 72 Cores per 2-Head Nodes
• Up to 48Gb Data Link from Head Nodes to RAID Storage Arrays
• Available in 16, 24, 48, 60 Drive Bays with up to 7 Expansion Units
• VSS Agents Support
11

• VMware VAAI Support
• VMware Storage Plug-in 5.5 and 6.0
• Storage can Provisioned as DataStores, Object (API Access), Block (iSCSI, Fibre),
or File (CIFS, NFS)
• NAS with Support for CIFS/SMB and NFS Protocols
• Unlimited NAS Nodes – Lab Tested up to 256 Nodes
• Support of Replica Nodes and RAID 10 Across All Nodes
• Scale Out NAS Snap
• Geo-Replication
• Cloud to Cloud Replication and Data Migration
• Native Linux Client for Fast Access to the Target File Node Location
• Object Access to File Storage, Filestore can also be Accessed using an Object-API
• Elastic Capacity – Stretch or Shrink Nodes on Demand with Auto Redistribution
• Bit-rot Detection to Help Preserve the Integrity of Data Assets by Detecting Silent
Corruption
• No Application Rewrites - Migrate Existing POSIX Applications to the Cloud with-
out Making Modifications
• Supports SELinux Enforcing Mode with SSL-Based In-Flight Encryption
StoneFly - The virtual computing platform for Software-Deﬁned
Data Center™
12

www.stonefly.com
2865, 2869 and 2879 Grove Way,
Castro Valley, CA 94546 USA.
+1 (510) 265-1616

Unveiling the Evolution: Proprietary Hardware to Agile Software-Defined Solutions

Recommended

More Related Content

Similar to Unveiling the Evolution: Proprietary Hardware to Agile Software-Defined Solutions (20)

More from MaryJWilliams2 (20)

Recently uploaded (20)

Unveiling the Evolution: Proprietary Hardware to Agile Software-Defined Solutions