Storage: Difference between revisions

From bwCloud-OS
Jump to navigation Jump to search
Admin (talk | contribs)
No edit summary
No edit summary
 
(73 intermediate revisions by 2 users not shown)
Line 1: Line 1:
<span id="In-a-Nutshell"></span>
{{InANutshell|
<li>If you need more storage in a VM, create and attach additional '''volumes''' — root disks cannot be resized. Using special flavors with larger root disks as a reserve for extra storage is generally discouraged, as they offer less flexibility. </li>
<li>All data is stored redundantly (e.g., 3× replication in Ceph), but '''no automatic backups''' are performed — ''you are responsible for backing up your own data''.</li>
<li>Back up your data regularly using '''snapshots''' or by downloading (volume-)images using the OpenStack client as explained
[[Storage#How_can_I_back_up_my_virtual_machines?|here]].</li>
<li>⚠️ ''To avoid data loss: '''Never store critical or irreplaceable data only in bwCloud-OS ''' — always back up externally''.<li>
}}




= Quickstart - Volumes/Storage =
__TOC__


= Volumes (Block Devices) =


{{:Quickstart: attach volume}}
== What should I do if the root disk of my VM becomes too small? ==
<span id="Volume-Attach"></span>


= Backups =


The size of the root disk (system disk) cannot be increased after the VM is created. If you need more storage, the recommended solution is to attach additional volumes (virtual disks) to your instance.


=== Bietet  bwCloud-OS eine besondere Schnittstelle für ein Backup / eine Datensicherung an? ===
If your current quota is not sufficient to create the needed volumes, please submit a ticket to [[Projects and Quota#Quota-Increase-Request|increase your volume quota]] via the [https://bw-support.scc.kit.edu bwSupportPortal].
Nein, wir bieten keine spezielle Schnittstelle an. Die Daten der (laufenden) Instanzen können aber mit herkömmlichen Bordmitteln gesichert werden.


The following guide walks you through all required operations.


=== Wie kann ich denn jetzt meine virtuellen Maschinen sichern? ===
[[Guide: Volumes and Images#Create and Attach a Volume|➡️ Volumes and Images Guide]]
Von Instanzen und angehängten Volumes können über das Dashboard Snapshots erzeugt werden. Snapshots von Instanzen sind Images und können mit den CLI Clients (openstack-client) heruntergeladen werden (Stichwort: ''openstack image download ...''). Von Volumes können ebenfalls Images erzeugt werden (Stichwort: ''create image from volume'') und auf dieselbe Art heruntergeladen werden. Bei großen Volumes ist es aber einfacher und effizienter, direkt aus den Instanzen heraus die Daten zu exportieren, z.B. mit rsync, scp o.ä. tools.


= Backups =


=== Volumes oder Images herunterladen ===
== Does bwCloud-OS provide a dedicated interface for backups? ==
Verbinden Sie sich mittel ''openstack-client'' mit der bwCloud. Erzeugen Sie aus Ihrem Volume ein Image.
<span id="Backup-Interface"></span>


<!--T:7-->
No, bwCloud-OS does not offer a dedicated interface for backups or data protection. However, you can back up your instance data using standard tools and methods, see below.
<pre>
# openstack volume list
# openstack image create \
    --volume <UUID> \
    my_volume_as_image
</pre>


== How can I back up my virtual machines? ==
<span id="Backup-VM"></span>


Image herunterladen:
You can back up your VMs in bwCloud-OS in several ways:
<pre>
# openstack image list
# openstack image save \
    --file my_image_file.img \
    <UUID>
</pre>


🔹 '''Snapshots via the Dashboard'''


Mit folgendem Befehl laden Sie eine lokale Image-Datei in den Abbildkatalog der gewählten Region hoch und erstellen dabei den Metadateneintrag. Das Image wird nicht in eine bestehende VM kopiert.
* You can create snapshots of both instances and volumes using the web interface.
* Instance snapshots are saved as images and can be downloaded using the CLI, see [[#Image-Export-Import|here]] for instructions.
* You can also convert volumes into images and download them the same way, see [[#Image-Export-Import|here]] for instructions.


🔹 '''Export data directly from the instance'''


<pre>
*For large volumes or more control, it's often easier and more efficient to back up data directly from inside the VM using standard tools such as <code>rsync</code>, <code>scp</code>, or other backup utilities.
# openstack image create \
    --property os_distro=linux \
    --property ssh_user=<USER> \
    --property hw_video_model=cirrus \
    --file my_image_file.img \
    <NAME>
</pre>


==Can I export or import a (volume-)image? ==
<span id="Image-Export-Import"></span>


=== Image in bwCloud-OS hochladen ===
Yes, that is possible. However, you should use the [[Programmatic_Access_and_Automation#OpenStack_Client|OpenStack CLI Client]]. The main commands can be found in the following guide.
Melden Sie sich im Dashboard an. Klicken Sie sich wie folgt durch die Grafische-Oberfläche:


[[Guide: Volumes and Images#Export and Import (Volume-)Images|➡️]] [[Guide: Volumes and Images#Exporting and Importing (Volume-)Images|Volumes and Images Guide]]


<pre> 'Compute' -> 'Abbilder' -> 'Abbild erstellen' </pre>
== How secure is my data (in terms of integrity) in bwCloud-OS? ==
= Performance =
<span id="Data-Integrity"></span>


== Drosselung des Datendurchsatzes == <!--T:13-->
Both the system disk (root disk) of a virtual machine and any attached volumes are stored redundantly, either in our distributed '''Ceph storage systems''' (most regions) or another redundant storage system (Freiburg). Each piece of data has redundancy level 3, which provides robust '''protection against individual hardware failures'''.


However, please note:


Aufgrund der internen Architektur liegen alle Daten (root discs der Instanzen, attached storage volumes etc.) im Ceph-Speicher der jeweiligen Region. Ceph ist ein netzwerkbasiertes verteiltes Speichersystem, was intern via Netzwerk mit den Compute-Hosts verbunden ist. Der verfügbare Speicherdurchsatz verteilt sich also auf alle aktiven Nutzer:innen. Hierbei gilt: Je mehr parallele Schreibvorgänge stattfinden, desto stärker sinkt der Datendurchsatz für jeden einzelnen. Was auch ganz logisch ist, denn die Netzwerkkapazität und die Bandbreite sind insgesamt limitiert.
*  '''No regular backups''' of VM data are performed. By default, neither the root disk nor attached volumes are backed up.
*  It is '''your responsibility''' to ensure that important data is backed up internally (e.g. via snapshots) and externally (using external backup systems).


=== ⚠️ Important Operational Note ===
bwCloud-OS is operated as a '''"best-effort" infrastructure service'''. This means:


Um möglichst allen Nutzer:innen eine gleiche Leistung anbieten zu können, wird der Speicherdurchsatz pro Instanz entweder auf 100 MByte pro Sekunde in beide Richtungen (full duplex) oder auf 800 IOPS begrenzt.
* To maintain performance and affordability, the system does not include mechanisms such as geo-redundancy.
* In very rare cases (e.g., multiple simultaneous disk failures or critical software bugs in Ceph), data loss may occur and recovery may not be possible.


== mehr Durchsatz erhalten == <!--T:16-->
''For this reason, we'' ''strongly recommend'' ''storing '''critical or irreplaceable data''' (including essential configuration files and recovery information) on '''external storage systems''' '''outside the bwCloud-OS'''.''


= Storage Performance =


Bei berechtigtem Anliegen können Nutzer:innen nach mehr Datendurchsatz anfragen. Bitte dazu ein [https://bw-support.scc.kit.edu/ Ticket] an uns schreiben. In dem Ticket teilen Sie uns folgende Punkte mit: 
== Is storage throughput in the bwCloud-OS throttled? ==
<span id="Throughput"></span>


Yes. Storage performance in bwCloud-OS is '''intentionally throttled''' to ensure fair usage across all users.


* Sie möchten mehr Datendurchsatz bekommen
All data — including root disks and attached volumes — is stored in the '''Ceph distributed storage system''', which is connected to the compute hosts via the internal network. This means that the storage performance depends on shared network resources''':''' The more simultaneous activity there is, the more the available bandwidth is divided among users.


To ensure fair access, each instance is limited to a '''maximum of 100 MB/s''' (full duplex) and '''800 IOPS.'''


* Beschreibung des Use-Case bzw. der Anwendung: Warum benötigen Sie mehr Datendurchsatz?
This limitation helps maintain consistent performance in a multi-user environment and prevents individual VMs from monopolizing bandwidth.


== Can I request higher storage throughput in bwCloud-OS? ==
<span id="Throughput-Increase"></span>


* OpenStack Kennung (idealerweise die User-ID)
Yes, in justified cases, higher data throughput can be granted.


Please submit a [https://bw-support.scc.kit.edu/ support ticket], including the following required information:


* In welcher Region benötigen Sie mehr Datendurchsatz?
* A short note that you are requesting '''higher storage throughput'''
 
* A description of your '''use case or application''' and why the increased throughput is needed
 
* Your '''OpenStack user ID'''
Die oben genannten Punkte sind Pflichtangaben.
* The '''region''' where the increased throughput is required

Latest revision as of 09:22, 22 October 2025

In a Nutshell
  • If you need more storage in a VM, create and attach additional volumes — root disks cannot be resized. Using special flavors with larger root disks as a reserve for extra storage is generally discouraged, as they offer less flexibility.
  • All data is stored redundantly (e.g., 3× replication in Ceph), but no automatic backups are performed — you are responsible for backing up your own data.
  • Back up your data regularly using snapshots or by downloading (volume-)images using the OpenStack client as explained here.
  • ⚠️ To avoid data loss: Never store critical or irreplaceable data only in bwCloud-OS — always back up externally.


Volumes (Block Devices)

What should I do if the root disk of my VM becomes too small?


The size of the root disk (system disk) cannot be increased after the VM is created. If you need more storage, the recommended solution is to attach additional volumes (virtual disks) to your instance.

If your current quota is not sufficient to create the needed volumes, please submit a ticket to increase your volume quota via the bwSupportPortal.

The following guide walks you through all required operations.

➡️ Volumes and Images Guide

Backups

Does bwCloud-OS provide a dedicated interface for backups?

No, bwCloud-OS does not offer a dedicated interface for backups or data protection. However, you can back up your instance data using standard tools and methods, see below.

How can I back up my virtual machines?

You can back up your VMs in bwCloud-OS in several ways:

🔹 Snapshots via the Dashboard

  • You can create snapshots of both instances and volumes using the web interface.
  • Instance snapshots are saved as images and can be downloaded using the CLI, see here for instructions.
  • You can also convert volumes into images and download them the same way, see here for instructions.

🔹 Export data directly from the instance

  • For large volumes or more control, it's often easier and more efficient to back up data directly from inside the VM using standard tools such as rsync, scp, or other backup utilities.

Can I export or import a (volume-)image?

Yes, that is possible. However, you should use the OpenStack CLI Client. The main commands can be found in the following guide.

➡️ Volumes and Images Guide

How secure is my data (in terms of integrity) in bwCloud-OS?

Both the system disk (root disk) of a virtual machine and any attached volumes are stored redundantly, either in our distributed Ceph storage systems (most regions) or another redundant storage system (Freiburg). Each piece of data has redundancy level 3, which provides robust protection against individual hardware failures.

However, please note:

  • No regular backups of VM data are performed. By default, neither the root disk nor attached volumes are backed up.
  • It is your responsibility to ensure that important data is backed up internally (e.g. via snapshots) and externally (using external backup systems).

⚠️ Important Operational Note

bwCloud-OS is operated as a "best-effort" infrastructure service. This means:

  • To maintain performance and affordability, the system does not include mechanisms such as geo-redundancy.
  • In very rare cases (e.g., multiple simultaneous disk failures or critical software bugs in Ceph), data loss may occur and recovery may not be possible.

For this reason, we strongly recommend storing critical or irreplaceable data (including essential configuration files and recovery information) on external storage systems outside the bwCloud-OS.

Storage Performance

Is storage throughput in the bwCloud-OS throttled?

Yes. Storage performance in bwCloud-OS is intentionally throttled to ensure fair usage across all users.

All data — including root disks and attached volumes — is stored in the Ceph distributed storage system, which is connected to the compute hosts via the internal network. This means that the storage performance depends on shared network resources: The more simultaneous activity there is, the more the available bandwidth is divided among users.

To ensure fair access, each instance is limited to a maximum of 100 MB/s (full duplex) and 800 IOPS.

This limitation helps maintain consistent performance in a multi-user environment and prevents individual VMs from monopolizing bandwidth.

Can I request higher storage throughput in bwCloud-OS?

Yes, in justified cases, higher data throughput can be granted.

Please submit a support ticket, including the following required information:

  • A short note that you are requesting higher storage throughput
  • A description of your use case or application and why the increased throughput is needed
  • Your OpenStack user ID
  • The region where the increased throughput is required