OpenStack Live Demo and Q&A (Video)

ozgurOpenStackLeave a Comment

On March 18 we hosted a Live Demo and Q&A of City Cloud OpenStack to let our viewers see all the key features of the system and ask all their questions to our developers. If you missed the Live show, here is the recording from the 60 minute session along with all the questions from the audience.

Most of the questions were answered live during the demo with more elaborate answers than we provide in plain text. Below the video you can find the specific demo sections and also the questions from the audience with time stamps for easier access to the parts that interest you the most.

Enjoy!

 

Demo Sections

  • Demo case brief @ 04:15
  • Creating servers @ 4:35
  • Server modifications @ 9:45
  • Creating a load balancer @ 12:20
  • Security Groups @ 18:35
  • Volumes @ 20:33
  • Cloning @ 23:23
  • Object Storage @ 27:32
  • Moving Volumes between data centers @ 30:40
  • Creating Volume Snapshots @ 33:45
  • Create a new volume based on a snapshot @ 34:10
  • Extending a volume @35:06
  • Orchestration via the OpenStack API @37:39
  • Modifying cores and/or RAM on a server @38:50
  • Consol access @40:28
  • Keypairs @44:00
  • API and logs @46:26
  • Load balancing types @57:15

 

Questions from the audience that were answered Live

[Answered Live at 14:20]
Q: So it is not possible to load balance servers from different data centers?
A: It is possible servers from anywhere. The load balancer does not really care what you load balance. It could be any server, or any other load balancer anywhere. Just enter the IP-address of the server you want to add to the load balancer and you are good to go. The Drag & Drop feature that we have implemented and that we are showing in the demo currently only works for servers located in the same data center as the load balancer.

[Answered Live at 15:40]
Q: Is this a hardware or software load balancer?
Q: So the load balancer will have to be its own machine or will you offer LBaaS?
A: It’s software load balancer service. (LBAAS = load balancer as a service)
A: It’s an LBAAS service. It will be shown shortly in the demo.

[Answered Live at 25:02] Q: What is the expected ETA for private images?
A: Q2 as the plan is. Can’t give you any set date but I would assume during april.

Q: what if you need an image that doesnt exist on your lists?
A: We will release “private images” that allows you to upload your own images. This is scheduled to be released in Q2.

Q: Can image backup be made and transferred to our local machines? Some customers requires local backups for security reasons.
A: Ones private images is released you will also be able to download a server

[Answered Live @ 32:20]
Q: Is it possible to have a volume mounted on several servers at once?
A: The volumes that exists today can only be attached to one server at a time. There is however a project in progress in openstack that aim to allow shared storage.

[Answered Live @ 35:58]
Q: will you talk about pricing?
A: Not in this webinar but our pricing is pretty straight forward and available at https://www.citycloud.com/pricing/

[Answered Live @ 35:25] Q: Are the networks transparent between the datacenters? (network subnet routable directly) or will i need to set up VPN connection between City DCs?
A: VPN as a service (VPNAAS) will be available in Q2. So you will be able to connect your networks between datacenters through that. And also connect a network to your office network or similar should you want to. We will also be able to do layer2 connectivity coming up in as a special service. More information will be available when that service is released.

[Answered Live @ 37:36] Q: Demo suggestions: Whats under the Orchestration menu? And What kind of API protocol do you use?
Q: How abaut Orchestration?
Q: When will the orchestration be available?
A: Orchestration is not available in the interface today. The API endpoints for orchestration are however available and can be used today.

[Answered Live @ 38:50] Q: Does server modifications like added memory, cores, etc require a reboot?
A: Yes the server is rebooted upon resize.

[Answered Live @ 40:28] Q: can you have console access to your server via the control panel?
A: Yes we have console access

[Answered Live @ 41:49] Q: Is it possible to move servers between projects.
A: Not at this point but it will be possible in the interface in Q2.

[Answered Live @ 42:53] Q: so with iptables rules we can also firewall access to the server.
A: Yes, you can control the access to your servers with security group rules.

[Answered Live @ 45:13]
Q: DO the server go offline while snapshot is created?
A: No it does not. It’s active and running during the snapshot.

[Answered Live @ 45:34]
Q: When will the database service be launched?
A: Hopefully in Q2 but it might be postponed to Q3.

[Answered Live @ 46:19]
Q: If I use the API, do you collect logs also?
A: If you are using the native OpenStack API you will NOT get any logs.

[Answered Live @ 48:03] Q: wich version of OpenStack are you using?
A: Latest stable, Juno – next version, Kilo, will be released 30th April, and we will as soon as possible with to kilo.

[Answered @ 48:32]
Q: can you access servers with ssh or is it only by controlpanel?
Yes, as long as the servers has a floating ip (public ip) they are accessible via SSH

[Answered Live @ 49:33] Q: Will it be possible to migrate generation 1 servers to openstack?
A: There is no such tool available today.

Q: Will i be able to migrate my “generation 1 servers” to openstack?
A: We are not offering any tools for this at the moment.

[Answered Live @ 50:13]
Q: Do you limit IOPS?
A: No we do not.

[Answered Live @ 50:51]
Q: Support for process monitoring?
A: We have the citymonitor service available in the control panel. We also offer monitoring services through our “managed hosting” or “on call” services. For more information about these services please contact our support.

[Answered Live @ 51:50] Q: can you loadbalance the loadbalancer between your different datacenters to avoid singlepoint of failure?

Q: since the lbaas is based on one datacenter there is a single point of failure if the datacenter goes offline? is there a setup that stays operational even if one datacenter goes offline ?
A: That is correct. The only way to handle this today is to handle where requests are sent. You can load balance between zones, but if a load balancer goes down you need to redirect traffic somehow. It might be a DNS with short TTL for instance. You can also use some other software for instance spinning up a server with PFSense in each datacenter.

[Answered Live @ 53:31] Q: Regarding the server toolbox icon… what is the difference between “Pause” and “Stop”? How does it affect costs?
A: The difference it the a STOP will set the server into state SHUTOFF and all resources except disk space will be released. PAUS will set the state to PAUSE and will NOT release any resources, and that being said, a stopped server will only be billed for the disk space used, but a paused server will be billed for all resources since the resources is still allocated.

[Answered Live @ 54:52] Q: Does the floating IP float between DCs?
A: No, Floating IPs are static to one DC at the moment, but we are looking at solutions to make this possible in the future.

[Answered Live @ 55:23] Q: is there any network throttling based on instance size ans is the interface speed shared with IO to the provisioned volumes?
No, all hardware setups have the same bandwidth (1Gbit)

[Answered Live @ 56:04]
Q: What is going to happen with generation1 VMs?
A: It will continue to exist side by side with the current system. If you have both services provisioned you can choose which “environment” to create your servers in.

[Answered Live @ 57:15] Q: what types of loadbalancing types exists? can i use the loadbalancer to send all traffic to first server and only send traffic to server 2 if 1 is down?
A: We are today offering to different algorithms, Round Robin and Least connections. The session persistency can be chosen between Source IP, HTTP Cookies and No session persistency

 

The rest of the questions and their answers

Q: During creation the disk storage selected is ephemeral storage local to provisioned machines or a connected one from SAN?
A: The local disk that you create at the moment you create the server is a local disk and can’t be moved to other servers. You can however create a volume based on the system disk (snapshot -> create volume from snapshot) and then attach that volume to another server.

Q: can I upload my own firewall and run it as a vm?
A: We will release “private images” that allows you to upload your own images. This is scheduled to be released in Q2.

Q: for remote administration we then need vpn access to the private network?
A: No you can manage servers through VNC console or through for instance RDP/SSH

Q: Please show how to balance both 80 and 443 to same members with the LBaas if that is possible.
A: Currently that is not possible. The load balancer works through a virtual IP with an assigned port number. However there are a lot of improvements coming in the load balancer in the next release (KILO release) which will be available at the end of April. Right now I don’t have a sure answer if it will be available in KILO release or not. We will have to look into that.

Q: When will it be possible to use both HTTP and HTTPS in the same loadbalancer?
A: Currently that is not possible. The load balancer works through a virtual IP with an assigned port number. However there are a lot of improvements coming in the load balancer in the next release (KILO release) which will be available at the end of April. Right now I don’t have a sure answer if it will be available in KILO release or not. We will have to look into that.

Q: why are the servers provided a private IP on creation? can i use any ip-range i want?
A: In our demo account we had a network/subnet and DHCP setup. (this is setup by default). This is fully configureable and you can choose to use static IPs instead.

Q: if you modify the size of a volume and try to recreate it from a snapshot from before the size change, how does the snapshot behave?
Q: ok, an example 🙂 the original volume is 100G, on which you do a snapshot. You then extende the volume to 150G, but later need to recreate it from the 100G snapshot, is the volume recreated as a 100G volume or left at 150G?
A: Openstack blocks if you try to extend a volume when there are snapshots attached to it. The reason for this is due to how snapshots are controlled on disk. When you do a snapshot the storage file created for the snapshot actually becomes the new active disk image which holds incremental changes made since the snapshot was created. The original image file (the volume itself) is left untouched. A bit difficult to explain without getting to technical, but basically this means that in order to resize a volume the made snapshots needs to first be merged into a single image file, which is done automatically when snapshots are deleted. So the scenario in the question can’t happen as the snapshots would first need to be deleted before the resize can be made.
Also note that when you create a volume from another volume or from a volume snapshot, you can choose the size of the new volume, but the size can not be smaller than the source.

Q: How is private networks isolated from eachother?
A: When you create private networks you can choose how they should be connected. You can connect them to the same router and allow traffic between them or you can keep them isolated if you want that. It’s all up to you.

Q: What do you recommend on the number of cores? I’ve heard that too many cores can be “hurtful” due to the scheduling on the physical cores.
A: It’s difficult to say what the ideal setup is. I’d say it depends on what you use the server for, what operating system it runs and what kind of services it will use. I haven’t had any case where more cores makes a system slower, however you might end up having more resources available than what the server can use if the software is not capable of taking advantage of all cores and RAM. All depending on what you run and how the server is configured.

Q: What VPN services are available (connecting to your own network)? IPsec, OpenVPN, others ?
Today we don’t have any build in OpenStack VPN (VPNaaS) – this will be released during Q2 2015, some time after the Kilo release is implemented. VPNaaS will be IPsec based when it is released.

Q: have you tested with Voip servers software?
Q: what about VOIP? have you tested with Voip Distributions?
A: No we have not

Q: rdp (=windows remote desktop protocol) access requires external ip?
Yes, you will need a public IP (floating ip)

Q: Storage space reservation options ? Eg reserve 10GB using 7200RPM disks or reserve 10GB using 15 000RPM disks?
A: We are offering our high speed storage for all services. Please read more at https://www.citycloud.com/cloud/city-cloud-demolishes-competition-disk-performance-results-techworlds-idg-big-test/

Q: What about DNS?
A: Today we DNS as a separate service, but will be included in the city control panel in the future

Q: Can IP address from Generation1 server be transferred/migrated into OpenStack server ?
A: No, IPs can not be shared between the different environments