The volume of data handled by many companies is growing rapidly, raising the question of where to store this data quickly, securely and cost-effectively. Instead of increasing their spending, in-house IT departments should focus on innovative technologies. And fortunately, online storage has now become commonplace.

The Open Telekom Cloud offers different types of storage, and prospective customers are well advised to make themselves familiar with the storage types that best fit their needs and budget beforehand. Not taking into account the separate topic of backups and disaster recovery, we will explain the three services Elastic Volume Storage (EVS), Object Storage Service (OBS) and Scalable File Service (SFS) in the following.

The various different storage media and their price/performance ratio play a crucial role in every storage scenario. Combining storage media in a hierarchy is referred to as “tiering.” Hard disks with SATA or SAS interfaces are by no means the fastest or cheapest storage medium, but they are most commonly used.

SSD (solid-state drive) storage media, which are available in several quality and performance levels, are considerably faster when it comes to data throughput. Thanks to their high performance, SSD and flash memory devices are enjoying widespread use, but their higher price is often only justified in analytics scenarios or in the field of e-commerce.

The slowest, but most reliable and cost-effective storage medium is magnetic tape. One tape cassette can now store around 200 TByte and because access is comparatively slow, tape is ideally suited for long-term, high-reliability storage. 

Block storage is measured in “volumes” (drives, folders), which is why Deutsche Telekom has chosen the term “Elastic Volume Service” (EVS). Block storage is important because each virtual machine (VM) needs to have a block storage. This storage pool can be assigned to various VMs (Shared EVS). With Shared EVS, it is possible to connect an EVS volume to up to 16 different VMs. 

Because block storage is connected via the network, speed plays a certain role during read/write processes. The volumes (storage units) are offered in five performance classes to meet different input/output speed requirements: SATA, SAS and SSD. In addition, there are performance-optimized SAS and SSD variants to enable higher data throughput (I/O), for example. Up to 32 TB of data can be stored on such a virtual hard disk, and up to ten such disks can be assigned to a VM. The system disks of the virtual machines are also based on EVS.

The Open Telekom Cloud offers block storage from three storage pools with different performance classes. The maximum size of a single virtual disk is 32 TB, and up to ten such disks can be attached to a VM. Each VM requires a basic hard disk for the operating system. Users can customize the storage capacity in real time.

SATA interfaces cover the requirements of normal data exchange rates (I/O rates), while SAS and SSD interfaces are used for higher requirements with redundant data storage. Multiple data storage offers particularly high data availability and data loss prevention,

and plays an important role in backup and disaster recovery scenarios. Redundant storage also offers particularly high I/O performance, with faster data exchange than when using object storage. Elastic Volume Storage can also be configured with an automatic backup. The user determines the parameters and frequency of the backups. The “EVS Encryption” feature offers the option of encrypting data in the block storage. 

EVS is available in three classes, and prices vary depending on access speed. The prices listed are monthly fees per GB provided. If the storage is only provided for half a month, you only pay half the monthly fee. As long as the storage has not been deleted, the storage capacity provided is also billed even if the instance (VM) connected to the Elastic Block Storage is stopped. 

Sample calculation

EVS is connected to the systems as required, the quantity varies from hour to hour. The quantity invoiced is then the average storage space used in the respective month – i.e. the sum of the individual hourly values divided by the number of hours per month.

In January, an average of 200 GB of storage space was allocated (148,800 GBh/744 h), which is the amount that will be invoiced. 

Object storage is the classic form of cloud storage. It is relatively cost-efficient and suitable for long-term data storage. Object storage works in the same way as teenagers clean up their rooms. Objects (images, videos, documents, etc.) are simply placed where there is space. In order to avoid ending up in a hopeless mess, object storage uses “buckets” to find the objects. Individual files have a maximum size of 50 TB. The Object Storage Service does away with conventional directory structures.

While this solution is cost-effective, it does not boast a particularly high performance. The objects cannot simply be modified directly in the storage, but they can only be uploaded, downloaded and deleted, and the buckets can be searched. To make a change to an object, it has to be downloaded, modified and then uploaded again.

In addition to pricing, object storage offers two additional advantages when compared to EVS.

1. It can also be used independently of computing resources (as a stand-alone system) and does not need to be connected to a VM. This means that it can be used to run static websites or to store web pages. A well-known example of this are operating instructions that users can access on a website as needed.
2. Object storage does not have to be expanded gradually as the amount of data increases, e.g. in an IoT deployment scenario. Object storage grows automatically as required, and is theoretically unlimited.

The Object Storage Service offers 99.99999999999% data stability and very high availability. The overall size of the object storage is scalable almost without limitations. Some vendors offer object storage in a variety of performance categories, typically three. The OBS is also available in three performance classes (cold, warm and standard), which are aimed at users who rarely access their data. This means that the storage model can be tailored to data usage in order to achieve the best possible price.

Cold OBS is ideal for access once a year at most, while warm OBS is geared towards monthly access. In the digital age, this is also a viable solution for test data that may still be used for subsequent big data analyses. This wealth of data can be stored in cold object storage until it is actually needed.

Cold object storage offers the lowest storage prices of all available solutions. Tapes (see above) are often used to achieve such low prices, but this also means that access to cold object storage is more expensive than to conventional “standard” object storage. Warm object storage is a compromise between standard and cold, although of course the object storage has to be integrated into an IT environment. 

The Open Telekom Cloud provides two common programming interfaces (“Swift” from OpenStack and Amazon S3), a web-based management console, a cloud storage web client and a software development kit (SDK) for development projects. The OBS is accessed via the standard HTTP/REST protocol layer, which can be used to delete, upload/download and search objects (see above). 

OBS supports data versioning to protect customer data even better against loss. This makes it easier for users to recover data they have accidentally deleted. Prior to storage, OBS also ensures that the data to be stored corresponds exactly to the data uploaded by the customer. It is stored in several segments so that sufficient redundancy can be achieved to ensure maximum availability. In the background, data consistency is automatically checked and corrupt files are restored. 

The Object Storage Service is billed on an average monthly basis according to the actual amount of data stored, e.g.: In the middle of the month, 4,000 GB are uploaded to the object storage. One request is allocated per uploaded file.

Because the data was uploaded in the middle of the month, the monthly average is 2,000 GB. The storage space used is charged for using a pricing scale.

Three positions are included in the invoice: The remaining quantity (up to 5 GB) and the next two increments (995 GB x costs of the 2nd increment + 1,000 GB x costs of the 3rd increment). 

With network storage, users gain the flexibility of object storage with higher performance – albeit with lower performance than when using real block storage. This is due to the fact that the storage is located somewhere else and has to be connected to the server via a network cable. On the other hand, it offers the added benefit of a convenient storage service.

Open Telekom Cloud’s Scalable File Service (SFS) is based on the OpenStack service Manila. The network storage is accessible across various availability zones from different virtual machines (VMs). With the NFSv3 protocol (NFS: Network File Storage), it uses a different technology than OBS and EVS and thus resembles a network-attached storage (NAS) device, i.e. mass storage that can be accessed via a network. This also means that several servers can access the storage unit. At the same time, it expands as required, all the way up to 10 PB. As the data volume grows, its network bandwidth automatically increases up to 2.5 Gbps. However, network storage offers higher performance than object storage.

The Scalable File Service is the most convenient storage option for some requirements, but above all the easiest to use. Users who are accustomed to Windows Explorer will find the overview very familiar. Just like the Object Storage Service (OBS), the Scalable File Service is charged for using a pricing scale. 

The Open Telekom Cloud offers a total of nine storage options. These options are based on different technologies and offer a wide range of different possibilities, for example when it comes to tiering. Before deciding on a storage option, it is important to determine the data requirements, and how often and for what purposes the data is to be used. This information enables users to find the optimum storage model that offers the best compromise between cost minimization and smooth operation.