Move faster with Network Controller Enhance network security































































- Slides: 63
Move faster with Network Controller Enhance network security VXLAN-based virtual networking Distributed firewall Hybrid SDN gateways for crosscloud deployment Network Security Groups for microsegmentation External and internal software load balancing Routing and mirroring to specialized virtual appliances Reduce costs Ability to converge RDMA and Ethernet traffic on the same teamed NICs Qo. S for predictable performance Monitoring and automation to reduce Op. Ex
“The ability to spin up a software-defined network in about eight minutes while eliminating a $20, 000 cost is a huge benefit. This task used to take over two weeks on average. ” Chris Amaaris Chief Technology Officer Convergent Computing
http: //software. intel. com/persistent-memory
https: //blogs. technet. microsoft. com/windowsserver/2 017/08/30/announcing-support-for-satadom-bootdrives-in-windows-server-2016/
VOLUMES STORAGE POOL
SMB 3 File Shares VOLUMES STORAGE POOL Scale-Out File Server
VM VM VM VOLUMES STORAGE POOL Hyper-Converged
VOLUMES STORAGE POOL SQL Server
Intel P 3700 800 GB NVMe SSD Vs Intel S 3700 800 GB SATA SSD $1, 299 $789 $1. 62 / GB $0. 98 / GB
Random 4 K IOPS / Node 95, 655 ? * 4 Nodes, Intel Xeon processor E 5 -2699 v 3, 10 Gb. E Networking, 100% 4 K Reads
Random 4 K IOPS / Node 95, 655 324, 018 * 4 Nodes, Intel Xeon processor E 5 -2699 v 3, 10 Gb. E Networking, 100% 4 K Reads
Intel P 3700 800 GB NVMe SSD Vs Intel S 3700 800 GB SATA SSD $1, 299 $789 $4. 00 / 1 K IOPS $8. 30 / 1 K IOPS
CPU / 100, 000 IOPS 37% ? * 4 Nodes, Intel Xeon processor E 5 -2699 v 3, 10 Gb. E Networking, 100% 4 K Reads
CPU / 100, 000 IOPS 37% 19% * 4 Nodes, Intel Xeon processor E 5 -2699 v 3, 10 Gb. E Networking, 100% 4 K Reads
IOPS per %CPU (70% Read, 30% Write, 4 k Random) Non-RDMA 12, 800 ? 4 x Servers, Intel Xeon E 5 -2699 V 4 (22 C/2. 2 GHz), 4 x 800 GB NVMe SSD, 2 x 40 Gbps Mellanox Connect. X-3 Pro
IOPS per %CPU (70% Read, 30% Write, 4 k Random) Non-RDMA 12, 800 RDMA 16, 300 4 x Servers, Intel Xeon E 5 -2699 V 4 (22 C/2. 2 GHz), 4 x 800 GB NVMe SSD, 2 x 40 Gbps Mellanox Connect. X-3 Pro
IO latency (90 th percentile write latency) Non-RDMA 390µs ? 4 x Servers, Intel Xeon E 5 -2699 V 4 (22 C/2. 2 GHz), 4 x 800 GB NVMe SSD, 2 x 40 Gbps Mellanox Connect. X-3 Pro
IO latency (90 th percentile write latency) Non-RDMA 390µs RDMA 250µs 4 x Servers, Intel Xeon E 5 -2699 V 4 (22 C/2. 2 GHz), 4 x 800 GB NVMe SSD, 2 x 40 Gbps Mellanox Connect. X-3 Pro
Windows Server Software Defined: The fastest route to the benefits of hyper-converged infrastructure Design Validate Deploy Learn more at microsoft. com/wssd
Trust a validated solution Accelerate time to value Deploy with confidence knowing your solution is partner-validated against our reference architecture. Get up and running quickly on a solution that's already built and tested by the partner of your choice. Optimize performance settings Take the guesswork out of solution tuning – offers are optimized to run best on each partner's hardware. Increase your hardware choices Select the vendor that's the best match for you, with new partners joining the program all the time.
New partners joining the program all the time
https: //docs. microsoft. com/enus/windows-server/sddc
Simplified evaluation Fast and easy to deploy Workload optimized All Intel® Select Solution configurations and benchmark results are verified by Intel Visit Intel. com/Select. Solutions for more information
Writes Integration with mirror-accelerated parity volumes: Deduplicated data automatically pinned to erasure coded tier Optional near-inline optimization on data rotation Mirror Erasure Coding Re. FS Data Rotation
SCM: Storage-Class Memory, Non-volatile storage connected via the CPU memory bus SSD: Any other Solid-State Drive connected via SATA or SAS SCM HDD: Any Hard Disk Drive connected via SATA or SAS NVMe: Non-Volatile Memory Express, connected via the PCIe bus (AIC, M. 2 and U. 2)
DRAM SCM SSD HDD 1 10
SCM as capacity devices OR SCM for Capacity OR SCM + NVMe for Capacity SCM + SSD for Capacity SCM as caching devices OR SCM for Cache NVMe for Capacity SCM for Cache SSD for Capacity
Scoped Volumes scoped to specific nodes Reduced impact of multi-node loss Marginal Drive Handling Hardening stack to detect drives reaching end of life or misbehaving ! Volume 1 Volume 2
A B’’ Server 1 A’ C Server 2 A’’ C’ Server 3 B C’’ Server 4 B’ Server 5
A B’’ A’ C’’ B Server 1 C Server 2 A’’ C’ B’ Server 3 Server 4 Server 5
Writes Improved performance with mirroraccelerated parity volumes: Faster data rotation Mirror Data compaction Up to 3 x faster Erasure Coding Re. FS Data Rotation
Project Kepler-47
Project Kepler-47 2 -Node proof-of-concept at Microsoft Ignite 2016 Low-cost hyper-converged with Storage Spaces Direct Great for remote/branch office Intel® Thunderbolt™ 3 networking Both nodes under 1. 0 cubic feet (or one carry-on bag)
$8, 995 for both - including drives 10 TB usable storage Intel® Xeon® E 3 -1230 v 6 3. 5 GHz Up to 60 K IOPS @ 20% CPU 64 GB DDR 4 EXX memory each Intel® Thunderbolt 3™ @ 20 Gb/s
Booth 1726
https: //docs. microsoft. com/en-us/windows-server/sddc
https: //myignite. microsoft. com/evaluations https: //aka. ms/ignite. mobileapp































































