The Appliance consists of a Linux-based 64-bit operating system and all the packages needed to run one of the server components of Nexthink: the Portal or the Engine. The Portal and the Engine must be installed in separate physical or virtual machines, except for some small setups, where they can share the same appliance. When installed in virtual machines, hardware requirements may vary depending on the load of the infrastructure.
Nexthink officially supports VMware and Hyper-V as virtualization platforms, but both the Portal and the Engine may run on any other virtualization platform of your choice. Beware that some versions of popular virtualization platforms may impose particular limits on the number of CPUs and amount of RAM that you can assign to a virtual machine. In installations with many devices, the possible maximum values may not reach the specified requirements. Likewise, in virtualized environments with high load, the performance of IO operations may not be sufficient for the Portal or the Engine to write to disk normally. In case of doubt, please contact Nexthink Customer Success Services to validate your virtualized setup.
In all cases, your servers must be powered by 64-bit compatible processors (AMD64 or Intel 64 -not Itanium- architecture). The vast majority of AMD and Intel processors currently available in the market comply with this requirement. Because of the high memory bandwidth demands of the Portal and the Engine, the installation on machines with a NUMA (Non-Unified Memory Access) architecture is not supported.
For installations on the cloud, the required types of Azure or AWS compute nodes are indicated.
The following definitions apply to all tables of requirements detailed below, for both virtual or physical appliances:
- CPU cores
- Number of CPU cores required by the Appliance. The reference model for a CPU core in a physical Appliance is a single CPU core of an Intel Xeon E5-2695 v3 @ 2.3 GHz. Fewer CPU cores may be required when using newer or faster CPUs in an Appliance. Likewise, depending on the measured performance of a specific setup and its particular CPU models, Nexthink may ask customers to increase the number of CPU cores in the Appliance to keep system usability up to acceptable levels. Always validate with Nexthink Customer Success Services in case of doubt.
- The amount of RAM required by the Appliance. As for the type of RAM, the minimal requirement for all configurations is DDR3-1600 with data rate of 1600 MT/s. Dedicate RAM exclusively to the Nexthink Appliance and avoid using shared memory, which may negatively impact the performance.
Many aspects of the usage of Nexthink and your infrastructure can affect the performance and requirements of your appliances: the total number of devices and their actual activity; the number of defined metrics, services, categories, scores, and remote actions; the type of hypervisor, the load of other VMs, the vCPU to pCPU ratio and the IOPS, when running on a virtual appliance. All of them have an impact on the performance and the amount of data that can be kept on the Nexthink appliances.
As the number of possible combinations is unmanageable, always validate your settings with the help of Nexthink Customer Success Services in case of doubt.
To help you size the Appliance for hosting the Portal, we define a metric called complexity. The complexity of a setup, along with the number of licensed devices, gives you an idea of the computation power required by the Portal for that particular setup:
complexity = entities * hierarchies * (max_levels + 2)
- entities is the total number of entities across all Engines;
- hierarchies is the total number of hierarchies;
- max_levels is the number of user-defined hierarchy levels for the hierarchy which has the largest number of levels (excluding both the root level and the entity level).
If you have already defined your hierarchies (in a pre-production environment, for instance), find these numbers in the Portal by logging in as administrator and navigating to ADMINISTRATION→Hierarchies:
Note the total number of entities (1) and the number of levels in the selected hierarchy (2) in the example above: 33 entities and 2 levels (Region and Country). If you have defined more than one hierarchy, select the hierarchy with the highest number of levels to use this number as value for max_levels in the formula.
According to the size and complexity of your setup, the hardware requirements of the Appliance that hosts the Portal are the following:
|Max devices||Max complexity||Memory||
Disk 500 metrics 1000 metrics
Details (90 days) 500 metrics 1000 metrics
|150 k||60 000||59 GB||
|100 k||40 000||41 GB||
|50 k||12 000||23 GB||
|20 k||4 000||17 GB||
|10 k||2000||13 GB||
|5 k||2 000||12 GB||
- Ask Nexthink Support for setups with more than 150k devices.
- The Portal requires at least 10 MB/s of disk throughput.
- The total number of entities across all Engines is limited to 8 000.
- The table shows the disk space required for a default maximum of 500 enabled metrics in green and for an absolute maximum of 1000 enabled metrics in red. The relation between disk space required and number is basically linear, so that doubling the number of metrics requires the double of disk space. Nexthink recommends to increase the default maximum number of enabled metrics gradually.
The quantities in the Details column correspond approximately to the additional disk space required to store 90 days of historical details of count metrics. Add the value in the Disk column to the value in the Details column to get the total disk space required. For more information, see the article about data retention in the Portal.
Once you have an Appliance that meets the requirements of the Portal, configure the Portal to allocate your hardware resources and make the most out of them.
The same hardware requirements stated above apply to Portals in virtualized environments. The recommended types of virtual machines for installing the Portal on the supported cloud platforms are the following:
|Max devices||Max complexity||Azure||AWS|
|150 k||60 000||Standard_E8s_V3||r5d.2xlarge|
|100 k||40 000||Standard_DS13_V2||r5d.2xlarge|
|50 k||12 000||Standard_DS4_v2||r5d.2xlarge|
|20 k||4 000||Standard_E4s_v3||r5.xlarge|
|5 k||2 000||Standard_DS3_V2||r5.large|
The following table holds the hardware requirements of the Engine for a few representative configurations:
|Max events||Max devices||Max entities||Memory||Disk||CPU cores||Network||Disk throughput|
|200 M||10 k||500||26 GB||250 GB||14||1 Gbps||SSD|
|100 M||10 k||100||18 GB||200 GB||10||1 Gbps||25 MB/s|
|50 M||3 k||100||8 GB||100 GB||5||100 Mbps||10 MB/s|
- Get in contact with Customer Success Services if you need to go over the limit of 200 million events.
- SSD drives are required for setups with more than 100 million events not only because of their high throughput but also because of their faster random accesses, which are critical for the correct functioning of the Engine. These are the reference specifications that your disk setup should match:
- Block size for single operation: 4 Kb
- IOPS: 20 000 (random access)
- Bandwidth: 80 MiB/s
- Latency: 0.1 ms
- If you install the Collector in servers, take into account for the sizing of the Engine that a single server is roughly equivalent to 20 normal devices.
- The indicated number of cores include 20 simultaneous Finder users. If more than 20 users access Nexthink Engine simultaneously, 1 additional core is required for each 5 users (up to a maximum of 24 cores).
- Tests under controlled conditions have demonstrated that the Engine is capable of dealing with up to 100 normalized Finder users when run on a 24 cores appliance (20 users for the first 8 cores + 16 cores * 5 users per core).
- A normalized user is characterized for querying the Engine once every 25 seconds with a query that takes 10% of a core dedicated to the Engine. If Finder users deviate too much from this behavior, the number of supported users may vary drastically. Note as well that any other kind of query to the Engine (such as queries to the Web API) reduces the number of supported users.
- The maximum number of supported mobile devices for all Engine configurations is 5 000.
- For Engines with Nexthink Act and more than 8000 devices, allocate an extra half GB of RAM (already accounted for in the table above).
Because memory consumption in the Engine highly depends on the usage of Nexthink, the most difficult hardware requirement of the Engine to adjust is the RAM. Once you have made a rough estimation of your hardware needs for the Engine, fine tune your RAM requirements by computing their value with the help of the following formula:
RAM [bytes] = #Events * 60 + #Devices * 500,000 + #Entities * #Services * 80,000 + 3 GB (system and proxy) + 0.5 GB (for Nexthink Act if #Devices > 8000)
Remember the maximum values per Engine for each parameter that you can enter in the formula:
#Events = 200 million #Devices = 10,000 #Entities = 500 #Services = 100
The same hardware requirements apply to Engines in virtualized environments. The following table completes the representative configurations in the table above with the recommendations for virtual and cloud platforms:
|Configuration||Virtual platform||Cloud platform|
|Max events||Max devices||MHz allocation to resource pool||Azure||AWS|
|200 M||10 k||from 1 to 2 Ghz per vCPU||Standard_F16s_v2||c5.4xlarge with SSD gp2|
|100 M||10 k||from 1 to 2 Ghz per vCPU||Standard_F16s_v2||c5.4xlarge with SSD gp2|
|50 M||3 k||from 1 to 2 Ghz per vCPU||Standard_F8s_v2||c5.2xlarge with HDD sc1|
The number of cores per socket must be one for vCPUS. Regarding the allocated MHz, it is fine to go with the lower value of the range if there is proper monitoring of the infrastructure in place. In case of performance issues, Support will ask you the monitoring information (CPU ready ratio, co-stop, RAM usage, ...) and will most probably ask you to increase your settings.
Regarding memory requirements, as the Engine is an in-memory database, it really depends on the way the hypervisor will address memory overcommitment with our appliance OS and Engine process. If the hypervisor can find and consolidate memory pages with identical content from the Engine VMs on the same host, it could be ok to overcommit. Again, in case of performance issues, Support will ask you the monitoring information about memory usage and overcommitment and may ask you to increase your settings.
Because the Portal mainly identifies the Engine appliances through the MAC address of their network cards for licensing purposes, it is important that the MAC address of your virtual appliances does not change with time. Use static assignment of MAC addresses on all your virtual appliances to avoid licensing issues, especially when rebooting the machines.
Running Nexthink on a single appliance
For very small installations, the Portal and the Engine can run on the same physical or virtual appliance.
|Max devices||1 000|
|Max complexity||2 000|
|Disk capacity||120 GB|
|Disk write speed||10 MB/s|
Nexthink traffic redirection service
The Collector traffic redirection service (nxredirect) is a tool included in the Engine appliance that resends activity information (UDP traffic) received from the Collectors to one or more additional Engines. Optionally, the redirection service is able to anonymize sensitive Collector data on the fly.
The hardware requirements of nxredirect depend on the service being run alongside the Engine or in an appliance where the Engine has been stopped:
- Nxredirect is run alongside the Engine
- The maximum number of supported devices is 10 000 without anonymization, or 5 000 if anonymization is switched on. The hardware requirements of the Engine apply (see table above). No additional hardware is needed.
- Nxredirect is run independently (i.e. the Engine has been stopped)
- The maximum number of supported devices depends heavily on anonymization being switched on or off, ranging from 5 000 up to 350 000 devices.
Assuming an average traffic per device as indicated in the product overview of the Collector, the hardware requirements of nxredirect are as follows:
|Max devices||Anonymization||CPU cores||Memory||Disk|
|Nxredirect + Engine||5 000||On||Engine reqs||Engine reqs||N/A|
|Nxredirect alone||5 000||On||2||5 GB||N/A|
The disk space requirements given for the Appliance already take into account the amount of space needed to keep up to ten internal backups of either the Portal or the Engine.
In the case that you activate external backups, Nexthink recommends you to reserve the following quantities of external storage, depending on the size of your setup. The figures indicate the file size for each individual backup.
The backup size for the Portal depends on the number of devices, the complexity, the amount of history and the number of widgets and reports. We recommend regularly monitoring the used capacity and adapting it based on actual needs.
|Max devices||External backup size|
|150 k||50 GB|
|100 k||30 GB|
|50 k||15 GB|
|20 k||10 GB|
|10 k||5 GB|
|5 k||3 GB|
The disk requirements for the backup of the Engine are more predictable than those of the Portal and only depend on the number of events stored in the Engine.
|Max events||External backup size||Network throughput|
|200 M||16 GB||30 MB/s|
|100 M||8 GB||25 MB/s|
|50 M||4 GB||15 MB/s|
To collect information from mobile devices synchronized via ActiveSync with Microsoft Exchange, the Mobile Bridge uses a Remote PowerShell connection to your Exchange Client Access server.
Install the Mobile Bridge on a dedicated Windows Server 2008 R2 or later. The hardware requirements for the Mobile Bridge are those same ones recommended by Microsoft for installing their operating system. The Mobile Bridge is compatible with Exchange 2010 SP2 or 2013.
|Without Web & Cloud||With Web & Cloud|
|Network card||Any, wireless or wired|
|Average network bandwidth||100-150 bps||150-250 bps|
Starting from Nexthink V6.3, the Finder supports high DPI screens. When setting DPI scaling in Windows, the Finder adapts its size properly.
|Memory||4 GB system memory, at least 2 GB available|
|Disk capacity||50 MB|
|CPU||2 cores, 2 GHz|
|Network||100 Mbps recommended|
Certified Hardware List
Nexthink V6 appliances include a Linux-based operating system that is derived from the freely distributed sources of a major North American Enterprise Linux vendor. This vendor maintains a list of supported hardware that has been tested and is certified to work with its Linux distribution. To help you choose your hardware for your appliances (the Portal and one or more Engines), verify that it is in the following list: