IOPS benchmark is very low
Pages: 1 2
Abhishek Kumar Singh
18 Posts
November 16, 2020, 6:12 amQuote from Abhishek Kumar Singh on November 16, 2020, 6:12 amWe have done PetaSAN clustering . Just need to know two things.
1. How to use HDD ,SSD & NVMe drives for best performance .
2. How to take backup non ceph options
Thank you
We have done PetaSAN clustering . Just need to know two things.
1. How to use HDD ,SSD & NVMe drives for best performance .
2. How to take backup non ceph options
Thank you
admin
2,974 Posts
November 16, 2020, 12:09 pmQuote from admin on November 16, 2020, 12:09 pm
- if your workload is virtualization, we recommend all flash. For non latency sensitive / slower iops you can use HDD witj SSD journal, you can also use SSD write cache to improve latency and iops iwith HDDs. Use NVMEss for journals to HDD ( ratio 1:12-16) or to SSD ( 1:4), using NVMEs for OSDs gives the same performance as SATA SSDs. Use enterprise grade SSD/NVME.
- We currently do not support backups but it is intended, we currently support replication among clusters. To support backups yourself you need to use Ceph directly to create snapshots and exporting images or snapshots differences, look at the rbd snap, rbd export and rbd export-diff commands, probably run in script or cron job.
- if your workload is virtualization, we recommend all flash. For non latency sensitive / slower iops you can use HDD witj SSD journal, you can also use SSD write cache to improve latency and iops iwith HDDs. Use NVMEss for journals to HDD ( ratio 1:12-16) or to SSD ( 1:4), using NVMEs for OSDs gives the same performance as SATA SSDs. Use enterprise grade SSD/NVME.
- We currently do not support backups but it is intended, we currently support replication among clusters. To support backups yourself you need to use Ceph directly to create snapshots and exporting images or snapshots differences, look at the rbd snap, rbd export and rbd export-diff commands, probably run in script or cron job.
Last edited on November 16, 2020, 12:10 pm by admin · #12
Abhishek Kumar Singh
18 Posts
November 20, 2020, 1:05 pmQuote from Abhishek Kumar Singh on November 20, 2020, 1:05 pm
Quote from admin on November 16, 2020, 12:09 pm
- We currently do not support backups but it is intended, we currently support replication among clusters. To support backups yourself you need to use Ceph directly to create snapshots and exporting images or snapshots differences, look at the rbd snap, rbd export and rbd export-diff commands, probably run in script or cron job.
Hi Team,
Can you please help me on commands to use ceph directly to create snapshots & exporting images or snapshots , rbd snap ,rbd export or script or cron jobs. In internet , We are not finding any useful blog about backup & restore .
Thank you in advance,
Abhishek
Quote from admin on November 16, 2020, 12:09 pm
- We currently do not support backups but it is intended, we currently support replication among clusters. To support backups yourself you need to use Ceph directly to create snapshots and exporting images or snapshots differences, look at the rbd snap, rbd export and rbd export-diff commands, probably run in script or cron job.
Hi Team,
Can you please help me on commands to use ceph directly to create snapshots & exporting images or snapshots , rbd snap ,rbd export or script or cron jobs. In internet , We are not finding any useful blog about backup & restore .
Thank you in advance,
Abhishek
admin
2,974 Posts
November 20, 2020, 10:01 pmQuote from admin on November 20, 2020, 10:01 pmIf you google for the commands i posted, you should get a lot of info.
there are also ready made tools
If you google for the commands i posted, you should get a lot of info.
there are also ready made tools
Abhishek Kumar Singh
18 Posts
November 25, 2020, 6:11 amQuote from Abhishek Kumar Singh on November 25, 2020, 6:11 amHow to create bond 2×1G for management
How to create bond 2×1G for management
admin
2,974 Posts
November 25, 2020, 8:07 amQuote from admin on November 25, 2020, 8:07 ameither from ui or manually if you edit the /opt/petasan/config/cluster_info.json
in ui when you deploy the first node and create a cluster, you can bond interfaces to the management interface. Since you are connected while doing this, make sure you choose a bond mode that does not require reconfiguration by the switch
either from ui or manually if you edit the /opt/petasan/config/cluster_info.json
in ui when you deploy the first node and create a cluster, you can bond interfaces to the management interface. Since you are connected while doing this, make sure you choose a bond mode that does not require reconfiguration by the switch
DividedByPi
32 Posts
November 25, 2020, 6:19 pmQuote from DividedByPi on November 25, 2020, 6:19 pmJust giving my honest opinion, if this cluster you are building is for any type of production environment, you should be looking at getting official support.. Just by the types of questions you are asking. I am not in any way trying to sound condescending, but many of your questions are very much intro questions. I would definitely just consider official support thats all!
Just giving my honest opinion, if this cluster you are building is for any type of production environment, you should be looking at getting official support.. Just by the types of questions you are asking. I am not in any way trying to sound condescending, but many of your questions are very much intro questions. I would definitely just consider official support thats all!
Pages: 1 2
IOPS benchmark is very low
Abhishek Kumar Singh
18 Posts
Quote from Abhishek Kumar Singh on November 16, 2020, 6:12 amWe have done PetaSAN clustering . Just need to know two things.
1. How to use HDD ,SSD & NVMe drives for best performance .
2. How to take backup non ceph options
Thank you
We have done PetaSAN clustering . Just need to know two things.
1. How to use HDD ,SSD & NVMe drives for best performance .
2. How to take backup non ceph options
Thank you
admin
2,974 Posts
Quote from admin on November 16, 2020, 12:09 pm
- if your workload is virtualization, we recommend all flash. For non latency sensitive / slower iops you can use HDD witj SSD journal, you can also use SSD write cache to improve latency and iops iwith HDDs. Use NVMEss for journals to HDD ( ratio 1:12-16) or to SSD ( 1:4), using NVMEs for OSDs gives the same performance as SATA SSDs. Use enterprise grade SSD/NVME.
- We currently do not support backups but it is intended, we currently support replication among clusters. To support backups yourself you need to use Ceph directly to create snapshots and exporting images or snapshots differences, look at the rbd snap, rbd export and rbd export-diff commands, probably run in script or cron job.
- if your workload is virtualization, we recommend all flash. For non latency sensitive / slower iops you can use HDD witj SSD journal, you can also use SSD write cache to improve latency and iops iwith HDDs. Use NVMEss for journals to HDD ( ratio 1:12-16) or to SSD ( 1:4), using NVMEs for OSDs gives the same performance as SATA SSDs. Use enterprise grade SSD/NVME.
- We currently do not support backups but it is intended, we currently support replication among clusters. To support backups yourself you need to use Ceph directly to create snapshots and exporting images or snapshots differences, look at the rbd snap, rbd export and rbd export-diff commands, probably run in script or cron job.
Abhishek Kumar Singh
18 Posts
Quote from Abhishek Kumar Singh on November 20, 2020, 1:05 pmQuote from admin on November 16, 2020, 12:09 pm
- We currently do not support backups but it is intended, we currently support replication among clusters. To support backups yourself you need to use Ceph directly to create snapshots and exporting images or snapshots differences, look at the rbd snap, rbd export and rbd export-diff commands, probably run in script or cron job.
Hi Team,
Can you please help me on commands to use ceph directly to create snapshots & exporting images or snapshots , rbd snap ,rbd export or script or cron jobs. In internet , We are not finding any useful blog about backup & restore .
Thank you in advance,
Abhishek
Quote from admin on November 16, 2020, 12:09 pm
- We currently do not support backups but it is intended, we currently support replication among clusters. To support backups yourself you need to use Ceph directly to create snapshots and exporting images or snapshots differences, look at the rbd snap, rbd export and rbd export-diff commands, probably run in script or cron job.
Hi Team,
Can you please help me on commands to use ceph directly to create snapshots & exporting images or snapshots , rbd snap ,rbd export or script or cron jobs. In internet , We are not finding any useful blog about backup & restore .
Thank you in advance,
Abhishek
admin
2,974 Posts
Quote from admin on November 20, 2020, 10:01 pmIf you google for the commands i posted, you should get a lot of info.
there are also ready made tools
If you google for the commands i posted, you should get a lot of info.
there are also ready made tools
Abhishek Kumar Singh
18 Posts
Quote from Abhishek Kumar Singh on November 25, 2020, 6:11 amHow to create bond 2×1G for management
How to create bond 2×1G for management
admin
2,974 Posts
Quote from admin on November 25, 2020, 8:07 ameither from ui or manually if you edit the /opt/petasan/config/cluster_info.json
in ui when you deploy the first node and create a cluster, you can bond interfaces to the management interface. Since you are connected while doing this, make sure you choose a bond mode that does not require reconfiguration by the switch
either from ui or manually if you edit the /opt/petasan/config/cluster_info.json
in ui when you deploy the first node and create a cluster, you can bond interfaces to the management interface. Since you are connected while doing this, make sure you choose a bond mode that does not require reconfiguration by the switch
DividedByPi
32 Posts
Quote from DividedByPi on November 25, 2020, 6:19 pmJust giving my honest opinion, if this cluster you are building is for any type of production environment, you should be looking at getting official support.. Just by the types of questions you are asking. I am not in any way trying to sound condescending, but many of your questions are very much intro questions. I would definitely just consider official support thats all!
Just giving my honest opinion, if this cluster you are building is for any type of production environment, you should be looking at getting official support.. Just by the types of questions you are asking. I am not in any way trying to sound condescending, but many of your questions are very much intro questions. I would definitely just consider official support thats all!