SERVERware Standalone
Introduction
SERVERware is a virtualization platform designed to host telephony and unified communications in the Cloud. This tool provides a wide range of IP services and applications with high availability, stability, and reliability for maximum uptime. SERVERware serves multiple types of customers with different PBXware editions at the same time, plus sipPROT, all through a single interface. The new graphical interface has made managing hosts, VPSs, and resources easier than ever.
Virtualization has become an essential pillar of modern IT infrastructure, offering unmatched flexibility, efficiency, and resource management for businesses of all sizes. At the heart of this technological revolution is SERVERware, an advanced virtualization solution developed by Bicom Systems. SERVERware paves the way for digital transformation and resource optimization by enabling businesses to consolidate their IT infrastructures, run multiple operating systems and applications on a single physical server, and do so efficiently and securely.
SERVERware provides shared, fault-tolerant, highly available and redundant network storage, high data integrity protection, and a built-in backup system.
Structure and requirements
Except for the Standalone edition, SERVERware will connect the disks of two storage servers into a single storage pool named “NETSTOR”. This is done using the ZFS file system as the file system and volume manager, combined with the iSCSI protocol. Storage size is limited only by physical constraints.
Connecting the disks to the storage requires a dedicated physical network link between the active and passive storage servers, with a minimum bandwidth of 1 Gb/s. This link is commonly referred to as the "RAN" (Replication Area Network). Latency is a critical element in this configuration, which implies that a direct cable connection between the servers is required, without any other intermediate network equipment such as switches or routers.
In this configuration, one of the servers is designated as the primary and provides storage for the processing hosts (Cluster Edition). Each local disk on the primary server is mirror-paired with a disk of equal or approximate size on the second server. In this arrangement, the disk on the second server is accessible via iSCSI and is exported to the RAN. In the event of a failure of the primary server, the second server is able to take over and continue to provide the service.
VPS
For each VPS, SERVERware creates a ZFS volume, and each volume is formatted with a different file system, usually ext4. In the case of using the Cluster Edition, these volumes are exported to the SAN (Storage Area Network) network corresponding to the appropriate processing host. The processing host attaches the volumes assigned to it and makes them accessible, then it runs the VPS.
The requirements for the SAN include the need for each server to have at least one 1 Gbit/s port, as well as a switch to connect the storage and processing hosts. To minimize latency between servers, it is recommended to keep this configuration as simple as possible. To ensure failover, the storage requires a floating IP address on the SAN network. This floating IP address will switch from the primary storage server to the secondary storage server in the event of a failover. It is essential to ensure that this feature is supported by the switch, taking into account any security settings (depending on the model).
To strengthen the reliability of the configuration when processing hosts are used, SERVERware supports bonding of network interfaces. It is possible to set up a redundant SAN network with two redundant switches. This means that two ports per server are aggregated into the SAN network. The same approach is possible for the RAN and LAN networks, but the SAN network is the most critical component and it is recommended to make it redundant before the others.
The installation of the storage configuration is automated and fully managed through the setup wizard. To automate this process, a complete topology must be in place, with servers installed and networks configured. In addition, accurate information about these networks is required to successfully complete this operation. In short, you will need to select the ports used for LAN, SAN and RAN, and choose the floating IP address, while specifying the location of the second storage server.
Scalability and storage maintenance
Storage scalability is limited by several factors, including the capacity of the RAN (Replication Area Network) and SAN (Storage Area Network) networks, as well as the number and size of disks. To increase storage capacity, it is possible to add more disks in pairs and configure the iSCSI targets according to the SERVERware edition.
For detailed information on expanding the storage pool, please refer to our documentation manuals.
Beware of Limitations: It is essential to note that no resource, whether storage, CPU, network, or memory, is unlimited. Therefore, careful management and use of these resources is crucial.
The following limits must be taken into account regarding storage:
- Space: The storage capacity depends on the number and size of the disks available in the network storage pool.
- Storage Speed: This speed is determined by the performance of the individual disks and the interface to which they are connected, such as SAS, SATA, RAID.
- SAN (Storage Area Network) capacity: This includes bandwidth and latency.
- RAN (Replication Area Network) capacity: This also includes bandwidth and latency.
It is important to consider space limitations as a ratio between actual usage and allocation. Not all VPS will use the entire storage space allocated to them, thus leaving room for overcommitting the available storage. However, it is crucial to remember that all VPS have the potential to reach maximum usage and request a storage expansion.
The amount of overcommitment possible will depend on the specific case and generally cannot be accurately predicted. It is therefore imperative to closely monitor the growth of storage usage and react accordingly. Careful planning of future storage, as well as well-defined scalability procedures, are essential to resolve issues related to space limitations.
As more and more VPSs are added or their usage increases, the demand for storage will also grow. To avoid potential problems, SERVERware continuously monitors the storage I/O load, and this information is displayed on the dashboard. In general, it is recommended that this I/O load never exceed 10%. If this limit is reached, it is necessary to take action to resolve this issue, such as creating a new cluster to offload the VPSs or upgrading the hardware for improved performance.
Bandwidth and latency limitations apply to SAN and RAN networks. It is therefore essential to observe certain considerations when planning and deploying VPS. The I/O bandwidth of each VPS can be limited, and it is crucial to use this option appropriately, taking into account the same considerations as for space overcommitment.
Not all VPSs will systematically use I/O at their maximum limit, but it remains possible. For VPSs with an unlimited I/O set, there is a potential risk when these VPSs use 100% of the SAN/RAN resources, which can affect the performance of all other VPSs and lead to various problems. In such cases, mitigation measures can be taken, such as upgrading the network infrastructure or creating more clusters to offload the affected VPSs. However, careful planning is required before deployment.
It is important to note that network and disk performance can vary and may be the source of bottlenecks. It is therefore essential, when troubleshooting, to determine precisely where the bottleneck lies in order to take appropriate steps to resolve it.
