For over a year and a half I have been working on building a custom NVMe Storage Server for my homelab. I wanted to build a high speed storage system similar to a NAS or SAN, backed with NVMe drives that provides iSCSI, NFS, and SMB Windows File Shares to my network.
The computers accessing the NVMe Storage Server would include VMware ESXi hosts, Raspberry Pi SBCs, and of course Windows Computers and Workstations.
The focus of this project is on high throughput (in the GB/sec) and IOPS.
The current plan for the storage environment is for video editing, as well as VDI VM storage. This can and will change as the project progresses.
More and more businesses are using all-flash NVMe and SSD based storage systems, so I figured there’s no reason why I can’t have build and have my own budget custom all NVMe flash NAS.
This is the story of how I built my own NVMe based Storage Server.
The first version of the NVMe Storage Server consisted of the IO-PEX40152 card with 4 x 2TB Sabrent Rocket 4 NVMe drives inside of an HPE Proliant DL360p Gen8 Server. The server was running ESXi with TrueNAS virtualized, and the PCIe card passed through to the TrueNAS VM.
The results were great, the performance was amazing, and both servers had access to the NFS export via 2 x 10Gb SFP+ networking.
There were three main problems with this setup:
Virtualized – Once a month I had an ESXi PSOD. This was either due to overheating of the IO-PEX40152 card because of modifications I made, or bugs with the DL360p servers and PCIe passthrough.
NFS instead of iSCSI – Because TrueNAS was virtualized inside of the host that was using it for storage, I had to use NFS since the host virtualizing TrueNAS would also be accessing the data on the TrueNAS VM. When shutting down the host, you need to shut down TrueNAS first. NFS disconnects are handled way healthier than iSCSI disconnects (which can cause corruption even if no files are being used).
CPU Cores maxed on data transfer – When doing initial testing, I was maxing out the CPU cores assigned to the TrueNAS VM because the data transfers were so high. I needed a CPU and setup that was better fit.
Version 1 went great, but you can see some things needed to be changed. I decided to go with a dedicated server, not virtualize TrueNAS, and go for a newer CPU with a higher Ghz speed.
And so, version 2 was born (built). Keep reading and scrolling for pictures!
On version 2 of the project, the hardware includes:
While the ML310e Gen8 v2 server is a cheap low entry server, it’s been a fantastic team member of my homelab.
HPE Dual 10G Port 560SFP+ adapters can be found brand new in unsealed boxes on eBay at very attractive prices. Using HPE Parts inside of HPE Servers, avoids the fans from spinning up fast.
The ML310e Gen8 v2 has some issues with passing through PCIe cards to ESXi. Works perfect when not passing through.
The new NVMe Storage Server
I decided to repurpose an HPE Proliant ML310e Gen8 v2 Server. This server was originally acting as my Nvidia Grid K1 VDI server, because it supported large PCIe cards. With the addition of my new AMD S7150 x2 hacked in/on to one of my DL360p Gen8’s, I no longer needed the GRID card in this server and decided to repurpose it.
I installed the IOCREST IO-PEX40152 card in to the PCIe 16x slot, with 4 x 2TB Sabrent Rocket 4 NVME drives.
While the server has a PCIe 16x wide slot, it only has an 8x bus going to the slot. This means we will have half the capable speed vs the true 16x slot. This however does not pose a problem because we’ll be maxing out the 10Gb NICs long before we max out the 8x bus speed.
I also installed an HPE Dual Port 560SFP+ NIC in to the second slot. This will allow a total of 2 x 10Gb network connections from the server to the Ubiquiti UniFi US-16-XG 10Gb network switch, the backbone of my network.
The Server also have 4 x Hot Swappable HD bays on the front. When configured in HBA mode (via the BIOS), these are accessible by TrueNAS and can be used. I plan on populating these with 4 x 4TB HPE MDL SATA Hot Swappable drives to act as a replication destination for the NVMe pool and/or slower magnetic long-term storage.
I may also try to give WD RED Pro drives a try, but I’m not sure if they will cause the fans to speed up on the server.
TrueNAS Installation and Configuration
For the initial Proof-Of-Concept for version 2, I decided to be quick and dirty and install it to a USB stick. I also waited until I installed TrueNAS on to the USB stick and completed basic configuration before installing the Quad NVMe PCIe card and 10Gb NIC. I’m using a USB 3.0 port on the back of the server for speed, as I can’t verify if the port on the motherboard is USB 2 or USB 3.
TrueNAS installation worked without any problems whatsoever on the ML310e. I configured the basic IP, time, accounts, and other generic settings. I then proceeded to install the PCIe cards (storage and networking).
All NVMe drives were recognized, along with the 2 HDDs I had in the front Hot-swap bays (sitting on an HP B120i Controller configured in HBA mode).
The 560SFP+ NIC also was detected without any issues and available to configure.
I created a striped pool (no redundancy) of all 4 x 2TB NVMe drives. This gave us around 8TB of usable high speed NVMe storage. I also created some datasets and a zVOL for iSCSI.
I chose to go with the defaults for compression to start with. I will be testing throughput and achievable speeds in the future. You should always test this in every and all custom environments as the results will always vary.
Initial configuration was done via the 1Gb NIC connection to my main LAN network. I had to change this as the 10Gb NIC will be directly connected to the network backbone and needs to access the LAN and Storage VLANs.
I went ahead and configured a VLAN Interface on VLAN 220 for the Storage network. Connections for iSCSI and NFS will be made on this network as all my ESXi servers have vmknics configured on this VLAN for storage. I also made sure to configure an MTU of 9000 for jumbo frames (packets) to increase performance. Remember that all hosts must have the same MTU to communicate.
Next up, I had to create another VLAN interface for the LAN network. This would be used for management, as well as to provide Windows File Share (SMB/Samba) access to the workstations on the network. We leave the MTU on this adapter as 1500 since that’s what my LAN network is using.
As a note, I had to delete the configuration for the existing management settings (don’t worry, it doesn’t take effect until you hit test) and configure the VLAN interface for my LANs VLAN and IP. I tested the settings, confirmed it was good, and it was all setup.
At this point, only the 10Gb NIC is now being used so I went ahead and disconnected the 1Gb network cable.
Sharing Setup and Configuration
It’s now time to configure the sharing protocols that will be used. As mentioned before, I plan on deploying iSCSI, NFS, and Windows File Shares (SMB/Samba).
iSCSI and NFS Configuration
Normally, for a VMware ESXi virtualization environment, I would always usually prefer iSCSI based storage, however I also wanted to configure NFS to test throughput of both with NVMe flash storage.
Earlier, I created the datasets for all my my NFS exports and a zVOL volume for iSCSI.
Note, that in order to take advantage of the VMware VAAI storage directives (enhancements), you must use a zVOL to present an iSCSI target to an ESXi host.
For NFS, you can simply create a dataset and then export it.
For iSCSI, you need to create a zVol and then configure the iSCSI Target settings and make it available.
SMB (Windows File Shares)
I needed to create a Windows File Share for file based storage from Windows computers. I plan on using the Windows File Share for high-speed storage of files for video editing.
Using the dataset I created earlier, I configured a Windows Share, user accounts, and tested accessing it. Works perfect!
Connecting the host
Connecting the ESXi hosts to the iSCSI targets and the NFS exports is done in the exact same way that you would with any other storage system, so I won’t be including details on that in this post.
We can clearly see the iSCSI target and NFS exports on the ESXi host.
To access Windows File Shares, we log on and map the network share like you would normally with any file server.
For testing, I moved (using Storage vMotion) my main VDI desktop to the new NVMe based iSCSI Target LUN on the NVMe Storage Server. After testing iSCSI, I then used Storage vMotion again to move it to the NFS datastore. Please see below for the NVMe storage server speed test results.
Note, that when I performed these tests, my CPU was maxed out and limiting the actual throughput. Even then, these are some fairly impressive speeds. Also, these tests were directly testing each NVMe drive individually.
Moving on to the NVMe Storage Server, I decided to test iSCSI NVMe throughput and NFS NVMe throughput.
I opened up CrystalDiskMark and started a generic test, running a 16GB test file a total of 6 times on my VDI VM sitting on the iSCSI NVMe LUN.
You can see some impressive speeds maxing out the 10Gb NIC with crazy performance of the NVME storage:
1145.28MB/sec WRITE (Maxing out the 10GB NIC)
62,725.10 IOPS READ
42,203.13 IOPS WRITE
Additionally, here’s a screenshot of the ix0 NIC on the TrueNAS system during the speed test benchmark: 1.12 GiB/s.
And remember this is with compression. I’m really excited to see how I can further tweak and optimize this, and also what increases will come with configuring iSCSI MPIO. I’m also going to try to increase the IOPS to get them closer to what each individual NVMe drive can do.
Now on to NFS, the results were horrible when moving the VM to the NFS Export.
You can see that the read speed was impressive, but the write speed was not. This is partly due to how writes are handled with NFS exports.
Clearly iSCSI is the best performing method for ESXi host connectivity to a TrueNAS based NVMe Storage Server. This works perfect because we’ll get the VAAI features (like being able to reclaim space).
iSCSI MPIO Speed Test
This is more of an update… I was finally able to connect, configure, and utilize the 2nd 10Gbe port on the 560SFP+ NIC. In my setup, both hosts and the TrueNAS storage server all have 2 connections to the switch, with 2 VLANs and 2 subnets dedicated to storage. Check out the before/after speed tests with enabling iSCSI MPIO.
As you can see I was able to essentially double my read speeds (again maxing out the networking layer), however you’ll notice that the write speeds maxed out at 1598MB/sec. I believe we’ve reached a limitation of the CPU, PCIe bus, or something else inside of the server. Note, that this is not a limitation of the Sabrent Rocket 4 NVME drives, or the IOCREST NVME PCIe card.
I’ve had this configuration running for around a week now with absolutely no issues, no crashes, and it’s been very stable.
Using a VDI VM on NVMe backed storage is lightning fast and I love the experience.
I plan on running like this for a little while to continue to test the stability of the environment before making more changes and expanding the configuration and usage.
Future Plans (and Configuration)
I plan to populate the 4 hot-swappable drive bays with HPE 4TB MDL drives. Configured with RaidZ1, this should give me around 12TB usable storage. I can use this for file storage, backups, replication, and more.
This design was focused on creating non-redundant extremely fast storage. Because I’m limited to a total of 4 NVMe disks in this design, I chose not to use RaidZ and striped the data. If one NVMe drive is lost, all data is lost.
I don’t plan on storing anything important, and at this point the storage is only being used for VDI VMs (which are backed up), and Video editing.
If I can populate the front drive bays, I can replicate the NVMe storage to the traditional HDD storage on a frequent basis to protect against failure to some level or degree.
Version 3 of the NVMe Storage Server
More NVMe and Bigger NVMe – I want more storage! I want to test different levels of RaidZ, and connect to the backbone at even faster speeds.
NVME Drives with PLP (Power Loss Prevention) for data security and protection.
Dual Power Supply
Let me know your thoughts and ideas on this setup!
A new Side Chat Episode of the Tech Informative is now live on YouTube. In this episode we are covering HPE Integrated Lights-Out, also known as HPE iLO.
The Tech Informative is a video podcast by Stephen Wagner and Rob Dalton that hopes to explore everyday technologies from the perspective of Information Technology professionals.
Rob Dalton is a lover of IT and a Director by profession. Rob considers himself a jack of all trades, an IT veteran, and is also the author of “Secured Packets”, a technology blog with a focus on security. Rob’s blog can be found at: https://www.securedpackets.com
In the ever-evolving world of IT and End User Computing (EUC), new technologies and solutions are constantly being developed to decrease costs, improve functionality, and help the business’ bottom line. In this pursuit, as far as end user computing goes, two technologies have emerged: Hosted Desktop Infrastructure (HDI), and Virtual Desktop Infrastructure (VDI). In this post I hope to explain the differences and compare the technologies.
We’re at a point where due to the low cost of backend server computing, performance, and storage, it doesn’t make sense to waste end user hardware and resources. By deploying thin clients, zero clients, or software clients, we can reduce the cost per user for workstations or desktop computers, and consolidate these on the backend side of things. By moving moving EUC to the data center (or server room), we can reduce power requirements, reduce hardware and licensing costs, and take advantage of some cool technologies thanks to the use of virtualization and/or Storage (SANs), snapshots, fancy provisioning, backup and disaster recovery, and others.
See below for the video, or read on for the blog post!
And it doesn’t stop there, utilizing these technologies minimizes the resources required and spent on managing, monitoring, and supporting end user computing. For businesses this is a significant reduction in costs, as well as downtime.
What is Hosted Desktop Infrastructure (HDI) and Virtual Desktop Infrastructure (VDI)
Many IT professionals still don’t fully understand the difference between HDI and VDI, but it’s as sample as this: Hosted Desktop Infrastructure runs natively on the bare metal (whether it’s a server, or SoC) and is controlled and provided by a provisioning server or connection broker, whereas Virtual Desktop Infrastructure virtualizes (like you’re accustomed to with servers) the desktops in a virtual environment and is controlled and provided via hypervisors running on the physical hardware.
Hosted Desktop Infrastructure (HDI)
As mentioned above, Hosted Desktop Infrastructure hosts the End User Computing sessions on bare metal hardware in your datacenter (on servers). A connection broker handles the connections from the thin clients, zero clients, or software clients to the bare metal allowing the end user to see the video display, and interact with the workstation instance via keyboard and mouse.
Remote Access capabilities
Reduction in EUC hardware and cost-savings
Simplifies IT Management and Support
Runs on bare metal hardware
Resources are dedicated and not shared, the user has full access to the hardware the instance runs on (CPU, Memory, GPU, etc)
Easily provide accelerated graphics to EUC instances without additional costs
Reduction in licensing as virtualization products don’t need to be used
Limited instance count to possible instances on hardware
Scaling out requires immediate purchase of hardware
Some virtualization features are not available since this solution doesn’t use virtualization
Additional backup strategy may need to be implemented separate from your virtualized infrastructure
If you require dedicated resources for end users and want to be as cost-effective as possible, HDI is a great candidate.
An example HDI deployment would utilize HPE Moonshot which is one of the main uses for HPE Moonshot 1500 chassis. HPE Moonshot allows you to provision up to 180 OS instances for each HPE Moonshot 1500 chassis.
Virtual Desktop Infrastructure virtualizes the end user operating system instances exactly how you virtualize your server infrastructure. In VMware environments, VMware Horizon View can provision, manage, and maintain the end user computing environments (virtual machines) to dynamically assign, distribute, manage, and broker sessions for users. The software product handles the connections and interaction between the virtualized workstation instances and the thin client, zero client, or software client.
Remote Access capabilities
Reduction in EUC hardware and cost-savings
Simplifies IT Management and Support
Runs as a virtual machine
Shared resources (you don’t waste hardware or resources as end users share the resources)
Easy to scale out (add more backend infrastructure as required, don’t need to “halt” scaling while waiting for equipment)
Can over-commit (over-provision)
Backup strategy is consistent with your virtualized infrastructure
Capabilities such as VMware DRS, VMware HA
Resources are not dedicated and are shared, users share the server resources (CPU, Memory, GPU, etc)
Extra licensing may be required
Extra licensing required for virtual accelerated graphics (GPU)
If you want to share a pool of resources, require high availability, and/or have dynamic requirements then virtualization would be the way to go. You can over commit resources while expanding and growing your environment without any discontinuation of services. With virtualization you also have access to technologies such as DRS, HA, and special Backup and DR capabilities.
Both technologies are great and have their own use cases depending on your business requirements. Make sure you research and weigh each of the options if you’re considering either technologies. Both are amazing technologies which will compliment and enhance your IT strategy.
This month on June 23rd, HPE is hosting their annual HPE Discover event. This year is a little bit different as COVID-19 has resulted in a change of the usual in-person event, and this year’s event is now being hosted as a virtual experience.
I expect it’ll be the same great content as they have every year, only difference is you’ll be able to virtually experience it from the comfort of your own home.
I’m especially excited to say that I’ve been invited to be special VIP Influencer for the event, so I’ll be posting some content on Twitter, LinkedIn, and of course generating some posts on my blog.
So you want to add NVMe storage capability to your HPE Proliant DL360p Gen8 (or other Proliant Gen8 server) and don’t know where to start? Well, I was in the same situation until recently. However, after much research, a little bit of spending, I now have 8TB of NVMe storage in my HPE DL360p Gen8 Server thanks to the IOCREST IO-PEX40152.
Unsupported you say? Well, there are some of us who like to live life dangerously, there is also those of us with really cool homelabs. I like to think I’m the latter.
PLEASE NOTE: This is not a supported configuration. You’re doing this at your own risk. Also, note that consumer/prosumer NVME SSDs do not have PLP (Power Loss Prevention) technology. You should always use supported configurations and enterprise grade NVME SSDs in production environments.
DISCLAIMER: If you attempt what I did in this post, you are doing it at your own risk. I won’t be held liable for any damages or issues.
NVMe Storage Server – Use Cases
There’s a number of reasons why you’d want to do this. Some of them include:
Virtualized Storage (SDS as example)
Special applications (database, high IO)
Adding NVMe capability
Well, after all that research I mentioned at the beginning of the post, I installed an IOCREST IO-PEX40152 inside of an HPE Proliant DL360p Gen8 to add NVMe capabilities to the server.
At first I was concerned about dimensions as technically the card did fit, but technically it didn’t. I bought it anyways, along with 4 X 2TB Sabrent Rocket 4 NVMe SSDs.
The end result?
IMPORTANT: Due to the airflow of the server, I highly recommend disconnecting and removing the fan built in to the IO-PEX40152. The DL360p server will create more than enough airflow and could cause the fan to spin up, generate electricity, and damage the card and NVME SSD.
Also, do not attempt to install the case cover, additional modification is required (see below).
Installing the card inside of the PCIe riser was easy, but snug. The metal heatsink actually comes in to contact with the metal on the PCIe riser.
You’ll notice how the card just barely fits inside of the 1U server. Some effort needs to be put in to get it installed properly.
There are ribbon cables (and plastic fittings) directly where the end of the card goes, so you need to gently push these down and push cables to the side where there’s a small amount of thin room available.
We can’t put the case back on… Yet!
Unfortunately, just when I thought I was in the clear, I realized the case of the server cannot be installed. The metal bracket and locking mechanism on the case cover needs the space where a portion of the heatsink goes. Attempting to install this will cause it to hit the card.
The above photo shows the locking mechanism protruding out of the case cover. This will hit the card (with the IOCREST IO-PEX40152 heatsink installed). If the heatsink is removed, the case might gently touch the card in it’s unlocked and recessed position, but from my measurements clears the card when locked fully and fully closed.
I had to come up with a temporary fix while I figure out what to do. Flip the lid and weight it down.
For stability and other tests, I simply put the case cover on upside down and weighed it down with weights. Cooling is working great and even under high load I haven’t seen the SSD’s go above 38 Celsius.
The plan moving forward was to remove the IO-PEX40152 heatsink, and install individual heatsinks on the NVME SSD as well as the PEX PCIe switch chip. This should clear up enough room for the case cover to be installed properly.
I went on to Amazon and purchased the following items:
They arrived within days with Amazon Prime. I started to install them.
And now we install it in the DL360p Gen8 PCIe riser and install it in to the server.
You’ll notice it’s a nice fit! I had to compress some of the heat conductive goo on the PFX chip heatsink as the heatsink was slightly too high by 1/16th of an inch. After doing this it fit nicely.
Also, note the one of the cable/ribbon connectors by the SAS connections. I re-routed on of the cables between the SAS connectors they could be folded and lay under the card instead of pushing straight up in to the end of the card.
As I mentioned above, the locking mechanism on the case cover may come in to contact with the bottom of the IOCREST card when it’s in the unlocked and recessed position. With this setup, do not unlock the case or open the case when the server is running/plugged in as it may short the board. I have confirmed when it’s closed and locked, it clears the card. To avoid “accidents” I may come up with a non-conductive cover for the chips it hits (to the left of the fan connector on the card in the image).
And with that, we’ve closed the case on this project…
One interesting thing to note is that the NVME SSD are running around 4-6 Celsius cooler post-modification with custom heatsinks than with the stock heatsink. I believe this is due to the awesome airflow achieved in the Proliant DL360 servers.
I’ve been running this configuration for 6 days now stress-testing and it’s been working great. With the server running VMware ESXi 6.5 U3, I am able to passthrough the individual NVME SSD to virtual machines. Best of all, installing this card did not cause the fans to spin up which is often the case when using non-HPE PCIe cards.
This is the perfect mod to add NVME storage to your server, or even try out technology like VMware vSAN. I have a number of cool projects coming up using this that I’m excited to share.
This link will take you to sign up for a HPE iLO Advanced trial license. After filling out the form, you’ll be able to download your iLO welcome letter, which includes your iLO key (that is valid through 2020), and instructions.
This is awesome, and will definitely help out a ton of IT administrators this year to remotely manage, monitor, and maintain their servers.
Since I’ve installed and configured my Nvidia GRID K1, I’ve been wanting to do a graphics quality demo video. I finally had some time to put a demo together.
I wanted to highlight what type of graphics can be achieved in a VDI environment. Even using an old Nvidia GRID K1 card, we can still achieve amazing graphical performance in a virtual desktop environment.
This demo outlines 3D accelerated graphics provided by vGPU.
Please see below for the video:
VMware Horizon View 7.8
NVidia GRID K1
GRID vGPU Profile: GRID K180q
HPE ML310e Gen8 V2
ESXi 6.5 U2
Virtual Desktop: Windows 10 Enterprise
Game: Steam – Counter-Strike Global Offensive (CS:GO)
Resolution of the Virtual Desktop is set to 1024×768
Blast Extreme is the protocol used
Graphics on game are set to max
Motion is smooth in person, screen recorder caused some jitter
This video was then edited on that VM using CyberLink PowerDirector
You may encounter a situation where you’re unable to connect to the management interface or NIC on your HPE MSA array. When this condition occurs, you are not able to ping the NIC, and the SMU (web interface) will not load.
When you visibly look at the array, the AMBER warning light may or may not be flashing.
If you have a dual controller setup, and connect to the SMU on the other controller, you may see numerous log entries where the management NIC port status changes repeatedly from up to down.
I’ve witnessed this issue occur on 2 separate HPE MSA 2040 storage arrays (both with dual controllers).
When you physically look at the management NICs on the controller in question, you’ll notice that the port status LED indicator turns on, and turns off repeatedly. The link status keeps changing from up to down (as reflected in the logs).
Restarting the unit will have no effect. Changing the network cable will have no effect.
To resolve this issue, you must play with the network cable and re-seat it a few times (possibly half-way if possible a couple times as sketchy as that sounds).
If you can get the link status up, and disconnect/reconnect the cable before the light turns off, the connection will stay up. It will continue to function and survive restarts until sometime in the future when you disconnect it and reconnect it.
Replacing the controller may also fix it, however in the first instance I observed this, the replacement controller exhibited the same behavior months later in the future.
Privacy & Cookies Policy