Skip to content

Network File Service (NFS)

Description

Network File Services (NFS) provide access to files over the network. We currently implements this service using EMC Isilon X410 systems, featuring:

  • 1.2 GB/s throughput per node
  • Low latency
  • SSDs for file system metadata and performance acceleration

We operate two clusters, each consisting of six individual nodes. All nodes within a cluster form a unified system. Storage capacity is provided as a global file system, ensuring that data is evenly distributed across all hard drives and nodes.

Up to two hard drives or an entire node can fail without affecting data availability. For every data set, ECC information (Reed-Solomon erasure coding) is simultaneously written across the entire cluster; this ensures that data can be recovered at any time after hard drive or node failures.

As a customer, you order shares, the size of which we can limit using quotas.

Service Objectives

  • Centralized file availability for multiple application servers
  • Provision of large storage volumes
  • Rapid recovery of virtual servers
  • Faster deployment
  • Cost savings compared to redundant local data storage
  • Prevention of incidents caused by insufficient local disk space on virtual servers
  • Reduction of identical change operations across virtual servers

What to Expect from Our NFS Service

  • High file availability for your setup
  • Fast access speeds
  • Up-to-date security patches for critical vulnerabilities
  • Virtually unlimited storage capacity
  • Daily backups per share
  • Daily snapshots based on a 7-day/4-week/3-month retention schedule

Optional Service Features

  • Data volume capping using hard and soft quotas to limit associated costs. If the soft quota is exceeded for seven consecutive days, access is restricted to read-only mode. The hard quota cannot be exceeded. Please note that this option may cause operational issues if write access is no longer possible.
  • Fixed 50 GB NFS share. If the limit is exceeded, the product is automatically upgraded to a dynamic 100 GB share.

Important Notes

The following points will help you avoid performance degradation. The closer you get to these maximum values, the more likely it is that your application will experience performance issues.

  1. Number of files in a directory: 10,000. We therefore recommend distributing files across subdirectories where necessary. This can be achieved, for example, using dynamic directory hashing.
  2. Number of subdirectories in a directory: 10,000
  3. File size: 1 TB

Technical Limits

  • Directory tree depth: 509 (see also directory path length)
  • Filename length: 63–255 bytes, depending on encoding. UTF-8 is the default.
  • Hard links per file: 1,000
  • Path length: 1,023 bytes

We reserve the right to modify the technical infrastructure and the associated performance and service parameters of the service at any time.

Fair Use Policy

We aim to avoid imposing hard limits on the use of shared infrastructure, in the interest of all customers. We want to accommodate unexpected peak loads in your setups. Please adhere to our recommendations. Should a customer's unusual use of shared infrastructure result in service degradation for other customers, we reserve the right to take measures to restore the intended performance level for the affected customers.