amazon Dell EMC PowerEdge R640 reviews
Overview of the Dell PowerEdge R640
The Dell PowerEdge R640 is an expandable 1U rack server designed for computing and storage via a 2-socket platform. Described as a balance between performance, cost and density, the Dell R640 is built to handle workloads from a variety of data center use cases – defined storage dense software, service providers, application layer, dense private cloud, virtualization and high-performance Computer (HPC). In addition, Dell has built the PowerEdge R640 as a server that can be easily deployed and can be continuously expanded from 3 to 1000 nodes for software-defined storage by using Dell EMC VxFlex Ready Nodes.
The Dell PowerEdge R640 contains a lot of powerful components, as well as lots of scalability. For example, it can be equipped with 2 Intel Xeon Scalable Family Processors with a maximum of 28 cores per processor. The Dell R640 has 24 DIMM slots for up to 3TB of RAM and up to 12 NVDIMM for up to 192GB of RAM. It also has two AC (or DC) backup power sources. Dell R640 also supports 3rd generation PCI expansion card.
Dell rack servers can be equipped with HDD (2.5in or 3.5in) and SSD, supporting up to 8 NVMe for those who are looking for the fastest storage performance possible. This number is double that of the Dell R630’s NVMe, so it certainly makes this 1U platform significantly more flexible. Finally, the Dell R640 can be configured with an 8 x 2.5 inch hard drive or 4 x 3.5 inch hard drive on the front or 10 x 2.5 inch hard drive on the front with optional support for the drive 2 X 2.5 inch hard on the back panel.
The PowerEdge R640 supports USB ports, NIC ports, VGA ports, serial connectors and IDSDM / vFlash cards that support optional flash memory cards and a dual SD module inside.
For testing, we configured the Dell R640 with Dual Intel Xeon Platinum 8180 processor and 384GB (32GB x 12) 2666MT / s. Storage data were achieved using 3.2TB NVME (2 x 1.6TB PM1725a NVMe SSD) and 2TB of SAS (SSD 5 x 400 GB PM1635a SAS).
Design and build
Although the Dell PowerEdge R640 server rack is built very compact, it still boasts a lot of flexibility as well as various configurations and options for expansion. Like we mentioned above, this includes an 8 x 2.5 inch drive system, 4 x 3.5 inch drive system and a 10 x 2.5 inch drive system for storage.
The control panel is located on the left side of the front, which monitors system health and system IDs, status LEDs and iDRAC Quick Sync 2 (wireless) indicators. The status LED shows any faulty hardware, while the wireless Quick Sync 2 indicates the system supports Quick Sync (a feature that allows administrators to manage the system via a mobile device).
As is the case with all rack servers, the large front part is designed for drive bays. In our 10-drive configuration, this means up to 10 hot-swap drives 2.5 inches (although users have the option to use 6 2.5 hot-swap drives) or up to 4 devices being NVMe.
Management tools
Like other PowerEdge Servers, the Dell R640 offers many management options.
Efficiency
In the evaluation of local system performance, we have the Dell R640 equipped with a powerful configuration with two different flash storage layers. The first is the NVMe flash, with 4 1.6TB SSDs and the second with SAS flash with 4 400GB SSDs. Both are Samsung drives, although specific parts may vary depending on which component is selected when the server is built. With Intel Platinum 8180 CPU inside, we have a lot of CPU cycles to solve our storage workload. As stated in our introduction, the server is equipped with dual Intel 8180 Platinum CPU as well as 384GB of RAM. For our application benchmark, we utilize ESXi 6.5.
SQL server performance
Our Microsoft SQL Server OLTP test protocol used is the draft of Transaction Processing Performance Council’s Benchmark C (TPC-C), an online transaction processing benchmark that simulates activities found in the environment complex applications. TPC-C benchmarks come closer than aggregate performance benchmarks to measure performance and bottlenecks of storage infrastructure in a database environment.
Each SQL Server server is configured with two vDisks: 100GB volume to boot and 500GB for databases and log files. From a system resource perspective, we have configured each VM with 16 vCPU, 64GB DRAM and using SCI LSI Logic SAS controller. While our Sysbench workload has previously tested saturated platforms in both I / O storage and capacity, SQL testing seeks performance latency.
This test uses SQL Server 2014 running on Windows Server 2012 R2 and Dell’s Benchmark Factory for Databases. Although the traditional usage of this benchmark is to test large 3,000-scale databases on local or shared storage, in this iteration, we focus on spreading the four bases 1,500-scale data on our servers.
Configuration of checking SQL server (per VM)
– Windows Server 2012 R2
– Storage Footprint: 600GB allocated, 500GB used
– SQL Server 2014
+ Database Size: 1,500 scale
+ Virtual Client Load: 15,000
+ RAM Buffer: 48GB
– Test Length: 3 hours
+ 2.5 hours preconditioning
+ 30 minutes sample period
For SQL Server, we evaluated individual VMs as well as aggregate points. Trading results showed that the aggregate score was 12,638.2 TPS with individual VMs with a range of 3,159.5 TPS to 3,159.6 TPS.
With the average latency of SQL Server, the Dell R640 has both an aggregate and individual VM latency of 4ms.
MySQL Sysbench performance
Our first local storage application benchmark includes Percona MySQL OLTP database measured through SysBench. This test measures average TPS (Transaction per second), average latency and average 99 percent delay as well.
Each Sysbench VM is configured with 3 vDisks: one for boot (~ 92GB), one with pre-built database (~ 447GB) and the third for the tested database (270GB). From a system resource perspective, we have configured each VM with 16 vCPU, 60GB DRAM and using SCI LSI Logic SAS controller.
– CentOS 6.3 64-bit
– Percona XtraDB 5.5.30-rel30.1
+ Database Tables: 100
+ Database Size: 10,000,000
+ Database Threads: 32
+ RAM Buffer: 24GB
– Test Length: 3 hours
+ 2 hours preconditioning 32 threads
+ 1 hour 32 threads
In our Sysbench standard, we tested the Dell R640 with the same layout as above. For trading performance, the server has an average aggregate TPS of 13,046 with individual VMs ranging from 3,231.4 TPS to 3,303 TPS.
For the average latency of Sysbench, the Dell R640 has a total score of 9.8ms with individual VMs running from 9.7ms to 9.9ms.
In our 99th worst percentile delay measurement, the server achieved an impressive composite score of 19.9ms with individual VMs between 19.7ms and 20ms.
where can you get a Dell EMC PowerEdge R640 online
Dell EMC PowerEdge R640 Server Bundle with 2X Silver 4114 2.2GHz 10C 32GB RAM H730P 2x240GB SSD (Renewed): Buy it now
Dell EMC PowerEdge R640 Server Bundle with 2X Bronze 3106 1.7GHz 8C 32GB RAM H730P 2x300GB (Renewed): Buy it now
Dell EMC PowerEdge R640 1U Rack Server – 1 x Xeon Silver 4110-16 GB RAM HDD – 240 GB (1 x 240 GB) SSD – 12Gb/s SAS, Serial ATA/600 Controller – 2 Processor Support – 3 TB RAM Support – Gigabit Ether: Buy it now
Dell EMC PowerEdge R640 Server Bundle with Gold 6126 2.6GHz 12C 32GB RAM H730P 2x240GB SSD (Certified Refurbished): Buy it now
Dell EMC PowerEdge R640 Server Bundle with 2X Gold 6148 2.4GHz 20C 64GB RAM H740P 2x120GB BOSS Card (Renewed): Buy it now
Dell EMC PowerEdge R640 Server Bundle with Silver 4114 2.2GHz 10C 32GB RAM H730P 2x240GB SSD (Certified Refurbished): Buy it now
Analyze VDBench’s workload
With the latest and greatest server, it will be interesting to launch the latest and greatest storage to get the biggest boost. However, not everyone will do this and some users will upgrade their servers with existing storage or with lower cost SAS-based flash drives. For review, we have included both NVMe and SAS storage for each benchmark. This is not a “better” scenario, because from a performance perspective, NVMe will win. The more a scenario “what to expect with certain storage” and should be evaluated this way.
Our final section on local performance testing focuses on aggregate workload performance. In this topic, we took advantage of 4 SAS and 4 NVMe SSDs in bare metal environments running Ubuntu 16.04.4. The workload is configured to emphasize 25% of each drive’s capacity, focusing on maintainability versus stable, worst-case performance.
When it comes to benchmark storage arrays, application testing is the best and synthetic testing is in second place. Although not a perfect representation for actual workloads, synthetic tests help basic storage devices with repeatability coefficients make it easier to compare between competing solutions. These workloads provide a variety of test configurations, ranging from “four-angle” tests, common database transfer size tests, as well as capture traces from environments VDI different. All of these tests take advantage of the popular vdBench workload generator, with a scripting tool to automate and obtain results through a large computing test cluster. This allows us to repeat the same amount of work on a variety of storage devices, including flash arrays and individual storage devices.
File:
– 4K Random Read: 100% Read, 128 threads, 0-120% iorate
– 4K Random Write: 100% Write, 64 threads, 0-120% iorate
– 64K Sequential Read: 100% Read, 16 threads, 0-120% iorate
– 64K Sequential Write: 100% Write, 8 threads, 0-120% iorate
– Synthetic Database: SQL and Oracle
– VDI Full Clone and Linked Clone Traces
Looking at the highest reading performance for SAS drives, the PowerEdge R640 can maintain a lag of less than one millisecond until it is near peak performance. The server broke 1ms at about 269K IOPS and peaked at about 271K IOPS with a delay of about 1.1ms.
For the top of NVMe that was read on the Dell R640, we found latency of less than one millisecond transparent with the highest performance of 2,711,968 IOPS with a delay of 186μs.
For the maximum write performance of the SAS, the R640 maintains a lag of less than one millisecond transparent with the highest performance of 266,641 IOPS and a latency of 807μs.
4K recording performance with peaked NVMe drives 1,265,764 IOPS with only latency of 191μs.
When we move on to consecutive benchmarks (64K), we’ll see the same kind of performance we saw with PowerEdge R7415. The start delay is high (19.8ms in this case) and decreases as the benchmark runs. R640 with SAS drives completed at 25,606 IOPS or 1.61GB / sec with a delay of 2.49ms.
For sequential readings of NVMe 64K, the R640 started to have very low latency and peaked at 193,493 IOPS or 12.1GB / s with a latency of 329μs.
Again, with sequential writing of 64K, the R640 with SAS started with a high latency (8,9ms) before ending at 27.394 IOPS or 1.71GB / s with a delay of 1.16ms.
Once again, the NVMe-based server started much lower with sequential recording and peaked at about 89K IOPS or 5.6GB / s with a lag of 315μs.
Turning to our SQL workload, SAS drives have a better overall display with a lag of less than one millisecond transparent, reaching a peak of 275,406 IOPS with a delay of 418μs.
For the SQL workload on the NVMe version of the server, we saw the highest performance of 930,251 IOPS with a delay of only 135μs.
For our SQL 90-10 with SAS, R640 has the highest performance of 268,036 IOPS with a latency of 448μs.
The NVMe version of SQL 90-10 has a server with the highest performance of 774,044 IOPS with a delay of 163μs.
SAS in SQL 80-20 has seen servers peak at 254.044 IOPS with a latency of 491μs.
For 80-20 SQL with NVMe, the R640 can peak at 652,259 IOPS with a delay of 193μs.
Turning to Oracle’s workload, the R640 due to SAS load peaked at 239,794 IOPS with a latency of 533μs.
For NVMe Oracle, the server peaked at 570,158 IOPS with a delay of 230μs.
For SAS Oracle 90-10, the server peaked at 263,745 IOPS and the delay was 327μs.
Oracle 90-10 with NVMe peaked at 615,818 IOPS with only latency of 141s .
Oracle 80-20 with SAS drives in the R640 gave us the highest performance of 239,107 IOPS with a delay of 361μs.
For Oracle 80-20 with NVMe drives, the server peaked at 535,046 IOPS with a delay of 163μs.
Next, we switched to VDI Replication Test, complete and linked. For VDI Full Clone Boot with SAS, the PowerEdge R640 peaked at 221,147 IOPS with a lag of 575μs before mitigating.
Looking at NVMe R640 when starting VDI Full Clone, the server peaked at 626,040 IOPS with latency 205s.
For the initial full copy of SAS VDI, the server has latency of less than one millisecond until about 105K IOPS and reaches a maximum of 107,280 IOPS with a delay of 1.11ms.
Initially log VDI Full Clone with NVMe with server peaks at 246,628 IOPS and 476μs for latency.
For the second VDI Full Clone Logging in with SAS, R640 drives peaked at 79,495 IOPS with 797μs latency.
With NVMe drives, the server can reach maximum at 161,771 IOPS and latency is 386s for second VDI Full Clone Logging.
Switch to the test Copy is linked to VDI, R650 loaded by SAS with the highest performance is 125,587 IOPS with 506μs latency.
For Clone Boot testing linked to NVMe VDI, the server has peaked at 346.693 IOPS and latency is 182s.
Initial login is linked with VDI with SAS indicating that the server peaked at 47,656 IOPS with a delay of 662μs.
For the initial log copy of NVMe, the R640 peaked at 87,384 IOPS with a lag of 359μs.
For the VDI’s second-linked Copy Log, the SAS-based PowerEdge R640 has a lag of less than one millisecond to about 59K IOPS and a maximum of 60,708 IOPS with a delay of 1.04ms.
And finally, the second VDI Linked Clone NVMe version Login has the server peak at 120,850 IOPS with a latency of 521μs.
Conclusion
Launched as one of the first 14th generation PowerEdge servers, the Dell EMC PowerEdge R640 is a 2-socket 1U server that balances energy, density and cost in a small footprint. The server can be set up with two Intel Xeon CPUs that can be extended to a maximum of 28 cores per processor and 24 DIMM slots that can hold up to 3TB of RAM or 12 can be filled with NVDIMM. For storage, users can equip the R640 with eight 2.5-tray, or four 3.5-tray, in the front (it can also be configured for 12 2.5 trays with 10 in the front and 2 in the back). Like all PowerEdge servers, the R640 will have a number of management options and tools including iDRAC and OpenManage. The server can handle a number of different use cases with Dell EMC pointing to SDS, service provider, application layer, dense private cloud, virtualization and HPC.
In our application performance benchmarks, we reviewed the performance of the PowerEdge R640 utilizing VMware to see individual VM performance as well as aggregate. In our SQL Server transaction test, we saw an aggregate score of 12,638.2 TPS and a combined delay of only 4ms. For Sysbench, we saw a combined trading performance of 13,046 TPS and an average delay of 9.8ms and for the worst case delay, we saw a aggregate lag of only 19.9ms.
Like its predecessors, the PowerEdge 1U family has a lot to offer, including a myriad of options and incredible frame level customization. This time, the Dell R640 offers a lot of configuration options. The Dell R640 is suitable for use with mainstream HCI such as vSAN / VxRail and XC series (Nutanix), as well as larger HCI / CI extensions such as VxRack SDDC. Of course, the R640 can work well in other classic environments outside the super-intense data centers like the traditional EMC can rely on software tools like OpenStack and Redfish. In any case, the Dell R640 is a great addition to the PowerEdge family and there are definitely many other uses that will be met.