Hyperconverged Infrastructure is supposed to simplify the data center by creating an environment that automatically scales as new applications and workloads are added to it. The problem is that the current generation of HCI solutions can only address specific use cases like virtual desktops or tier 2 applications. First generation HCI solutions don’t have the per node power to accommodate enterprise workloads and tier 1 applications. The organization needs a next generation HCI solution, HCI 2.0, that can address HCI 1.0 shortcomings and fulfill the original promises of HCI; Lower costs, faster innovation, simpler scale, single vendor and unified management. The combination enables HCI 2.0 to handle a variety of storage intensive workloads.
3. The Components of Hyperconverged
● Server Nodes, NICs, Drives
● Network Switches
● HCI Software Stack
○ Hypervisor software
○ Storage software running on each
server node
○ Networking software running on
each server node
4. Goals of Hyperconverged Infrastructure
● Provide one platform for compute,
storage and networking
● Simplify scaling and management
● Lower costs
○ Industry standard components
5. The State of Hyperconverged
Infrastructure (HCI 1.0)
● Data Centers typically use HCI for one or two
use cases
○ Another environment to manage, not a
single environment
○ Not used for Tier 1 workloads
● Lack of per node computing, storage
performance and capacity leading to node
sprawl
● Node sprawl eliminates price advantages
7. First, Why
Scale? ● Four Reasons to Scale
○ More compute to support more VMs
○ More storage performance
○ More storage capacity
○ Capacity typically outpaces
performance
8. ● Traditional server nodes lack per node
○ Compute
○ Storage Performance
○ Storage Capacity
● Most HCI 1.0 Expansions are for capacity
○ CPU performance goes largely unutilized
Second, Why Does HCI 1.0 Scale-Out So Quickly?
9. Third, Impact
of Node
Sprawl
● Server costs go up
○ Eliminates the advantage of commodity nodes
● Networking costs go up
○ More ports are consumed by more nodes
● Licensing costs go up
○ More sockets means higher pricing
● Operational expenses go up
○ High node count clusters are complex to
manage
11. What is
HCI 2.0? ● High performance and high capacity
server nodes
○ High CPU core count per node
■ Enough CPU cores to run hundreds
of VMs
○ High storage capacity per node
■ High device count per node
○ High storage performance per node
■ NVMe flash in each node
12. Impact of HCI 2.0
● Much lower server node count per
cluster than 1.0
○ Scales-in before scaling out
● Lower Cost
○ Less server nodes
○ Less networking
○ Better software licensing efficiency
● High Performance for:
○ Tier 1 applications
○ High VM to physical server ratios
13. About Axellio Inc.
axellio.com
X-IO Technologies
manufactured:
Sold Storage Array
Business:
Axellio will focus on the Edge
Compute environment in
three areas:
• HCI 2.0 - HCI for Storage
Intensive Workloads
• Edge MicroCloud / Micro
Data Center
• Packet Capture and
Playback (security, signal
processing, tick analytics
etc.)
• Shared Storage Arrays
since 1997
• Storage Intensive
Servers since 2017
• X-IO Storage sold to
Violin Systems for
cash and stock
• X-IO Technologies
becomes Axellio Inc.
15. www.axellio.com
Why Hardware Matters in
Hyper-Converged Infrastructure
14
axellio.com
PCIe PCIePCIe PCIe
PCIe
Storage Controller
Up to 24 SAS/SATA
SSD
4 - 16x
PCIe
PCIe PCIePCIe PCIe
PCIe PCIe
PCIe PCIe
2-24NVMeSSD
2-24NVMeSSD
2-24NVMeSSD
2-24NVMeSSD
2-24NVMeSSD
2-24NVMeSSD
32x PCIe
PCIe PCIePCIe PCIe
PCIe
Storage
Controller
Up to 24
SAS/SATA
SSD
4 - 16x
PCIe
PCIe PCIe
1NVMeSSD
1NVMeSSD
Server with NVMeServer FabricXpress
16. The Value of 3x More
NVMe SSDs per Server
• The highest number of NVMe SSDs available in a server from any other
manufacturer is 24.
• Why is having 3x more SSDs on 3x more PCIe lanes per server so much better?
It’s about:
• Workload carrying capacity and performance:
• I/O sustained bandwidth
• Predictable low I/O latency
• Smaller clusters of larger servers decreases costs:
• More SSDs per server node allows for larger CPUs to be fully utilized
• Reduces socket based licensing costs
Enables you to move tier 1, storage intensive application workloads onto HCI to get
the benefits of simplicity and far lower total cost of ownership
You can now “Can” your SAN!
axellio.com
20. Why Consider FX-HCI 2.0 for Your Environment
Smaller clusters of Axellio servers offer multiple benefits:
• Lower latency across your applications
• Ability to support a multi-tenant application environment
• Decrease software licensing cost on a per server node basis – Oracle, SQL, etc…
Traditional servers can’t support the enterprise workload demands AND provide simplicity –
FX-HCI 2.0 CAN
• IT sprawl is experienced with traditional servers in HCI 1.0 and 3-2-1 SAN environments
• With FX-HCI 2.0 – add storage independent of compute to the greatest number of NVMe devices in a server than
available through ANY other server provider
FX-HCI offers:
• Software stack that meets your need (You may already own it!!)
• Installation and migration services - Easily migration from any environment to FX-HCI
• Support – Like you would expect from an enterprise storage company
• Feature Rich – All the features you have come to expect from an enterprise infrastructure solution
axellio.com
Power Your Enterprise Workload Demands AND Decrease Complexity &
Costs with Hyper-Converged Infrastructure 2.0