<div dir="ltr"><div><div class="gmail_extra"><div class="gmail_quote">On Mon, Mar 27, 2017 at 8:59 AM, Tomas Jelinek <span dir="ltr"><<a href="mailto:tjelinek@redhat.com" target="_blank">tjelinek@redhat.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><br><div class="gmail_extra"><br><div class="gmail_quote"><div><div class="gmail-m_-971199036686354921h5">On Mon, Mar 27, 2017 at 1:32 PM, Juan Hernández <span dir="ltr"><<a href="mailto:jhernand@redhat.com" target="_blank">jhernand@redhat.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><span class="gmail-m_-971199036686354921m_4462347545091815787gmail-">On 03/27/2017 01:03 PM, Tomas Jelinek wrote:<br>
><br>
><br>
> On Mon, Mar 27, 2017 at 11:21 AM, Juan Hernández <<a href="mailto:jhernand@redhat.com" target="_blank">jhernand@redhat.com</a><br>
</span><div><div class="gmail-m_-971199036686354921m_4462347545091815787gmail-h5">> <mailto:<a href="mailto:jhernand@redhat.com" target="_blank">jhernand@redhat.com</a>>> wrote:<br>
><br>
> Top posting, sorry.<br>
><br>
> There are a few things I'd like to clarify, regarding this subject:<br>
><br>
> 1. Data aggregation, as requested now by Tomas, and by other people in<br>
> the past.<br>
><br>
> We used to have that 'detail' parameter, to aggregate certain very<br>
> specific types of data, in particular to aggregate VM disks and NICs. We<br>
> removed that in version 4 of the API because the implementation was<br>
> extremely inefficient, from the engine point of view. An innocent<br>
> request like this:<br>
><br>
> GET /ovirt-engine/api/vms?detail=+<wbr>disks,+nics<br>
><br>
> Would generate, with the implementation we used to have, 1 query for the<br>
> VMs and then as many queries for disks and NICs as VMs in the system. In<br>
> our scale test environments, for example, with approx 4000 VMs and 10000<br>
> disks, that would take more than 20 hours to execute.<br>
><br>
> In addition, we didn't have in the past any mechanism to make this<br>
> available in a generic one, because there was no knowledge in the API of<br>
> what are 'details'.<br>
><br>
> In version 4 of the API we introduced a formal (kind of) specification<br>
> of the API (a.k.a. the model), and int includes knowledge about what are<br>
> 'links'. For example, the specification of the VM type contains this:<br>
><br>
> @Link DiskAttachment[] diskAttachments();<br>
> @Link Nic[] nics();<br>
><br>
> With this information we are now in a position where we can implement<br>
> this in a generic way.<br>
><br>
> We intend to implement this using a mechanism similar to the existing<br>
> 'detail' parameter:<br>
><br>
> GET /ovirt-engine/api/vms/123?foll<wbr>ow=disk_attachments,nics<br>
><br>
> The naive implementation of this is to let the API call itself. For<br>
> example, when the user requests to follow the 'disk_attachments' detail<br>
> the API can just call itself to get that:<br>
><br>
> GET /ovirt-engine/api/vms/123/disk<wbr>_attachments<br>
><br>
> However, we can't use that naive approach, if we do we end with the<br>
> 1+C*N query problem described before. We need to use specific<br>
> implementations for certain frequent use cases, like VMs+disks+nics, and<br>
> that needs work in the API and in the backend.<br>
><br>
> Tomas, if you want to help moving this forward, please open a RFE and<br>
> makes sure it gets attention.<br></div></div></blockquote><div><br></div></div></div><div>ok, opened: <a href="https://bugzilla.redhat.com/show_bug.cgi?id=1436206" target="_blank">https://bugzilla.redhat.com/sh<wbr>ow_bug.cgi?id=1436206</a></div><div>Will try to get it done soon.</div></div></div></div></blockquote><div><br></div><div>Forgive me if this is radical, but has anyone thought of / discussed using a NoSQL alternative to our very normalized SQL db as a way to avoid the problem of aggregating details? Using mongodb as an example, embed some of the smaller objects, and there's no cost of aggregation there. IIRC, Doctor REST uses mongo under the hood.<div><br></div><div><a href="https://docs.mongodb.com/manual/tutorial/model-embedded-one-to-many-relationships-between-documents/">https://docs.mongodb.com/manual/tutorial/model-embedded-one-to-many-relationships-between-documents/</a><br></div><div><br></div><div>Greg</div></div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><div class="gmail_extra"><div class="gmail_quote"><div><div class="gmail-m_-971199036686354921h5"><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div><div class="gmail-m_-971199036686354921m_4462347545091815787gmail-h5">
><br>
><br>
> This sounds pretty good! I will open, but since we are talking already<br>
> here I'll just use the opportunity to clarify the topic more and than<br>
> I'll open the BZ.<br>
><br>
> What I can imagine is the GetAllVmsQuery will accept in params also the<br>
> list of details it should provide. Than, the GetAllVmsQuery will<br>
> implement the efficient way of retrieving this info as well.<br>
><br>
> So, from the API perspective, it will be about taking the<br>
> ?follow=<something> part and passing it to the backend query params.<br>
><br>
> What you think?<br>
><br>
<br>
</div></div>Exactly, that is the point! The API by itself can't optimize database<br>
queries, all it can do is call the backend. It is the backend that has<br>
the opportunity and possibility to send optimized queries to the database.<br>
<br>
For other less common things we can use the naive approach, and<br>
implement the aggregation in the API itself. But for common use cases,<br>
like VM+disks+nics, we need to do it in an efficient way.<br>
<div><div class="gmail-m_-971199036686354921m_4462347545091815787gmail-h5"><br>
><br>
><br>
> 2. Reuse of TLS sessions.<br>
><br>
> The part of creating TLS sessions that is expensive is the generation of<br>
> the shared session key. That can be avoided if both the server and the<br>
> client are careful and reuse the session, using the session cache<br>
> mechanism built-in into TLS itself. The web servers that we use (Apache<br>
> and Undertow) do implement this mechanism, and so do most of our<br>
> clients. Make sure that your client uses it as well. In Java this is<br>
> achieved re-using the SSLContext. We already do that for the engine to<br>
> VDSM communciation for example. In JavaScript the browser already takes<br>
> care of this.<br>
><br>
> 3. Parallelism and latency.<br>
><br>
> A typical problem that we have is that we send many request to the<br>
> server. For example, to retrieve user sessions for a set of VMs we tend<br>
> to send many requests like this:<br>
><br>
> GET /ovirt-engine/api/vms/1/sessio<wbr>ns<br>
> GET /ovirt/engine/api/vms/2/sessio<wbr>ns<br>
> GET /ovirt-engine/api/vms/3/sessio<wbr>ns<br>
> ...<br>
><br>
> And we do that in a synchronous way: send one, wait for the result, send<br>
> another one, wait for the result, etc. This means that we don't take<br>
> advantage of the parallelism of the server and that we add to each<br>
> request the network round trip time. So if we have N requests, we have<br>
> to wait at least N*RTT.<br>
><br>
> The web servers that we use support multiple connections, and the<br>
> protocol that we use, HTTP, supports pipe-lining. This means that you<br>
> can send multiple requests in parallel, and that you can send multiple<br>
> requests without waiting for the response. To give you an idea of the<br>
> improvement that can be achieved, we recently added asynchronous request<br>
> support to the Ruby SDK, with multiple connections and pipe-lining. In<br>
> our scale testing environment that reduced the time to collect a<br>
> complete inventory from approx 30 min to approx 2 min. Here you have an<br>
> example:<br>
><br>
><br>
> <a href="https://github.com/oVirt/ovirt-engine-sdk-ruby/blob/master/sdk/examples/asynchronous_inventory.rb" rel="noreferrer" target="_blank">https://github.com/oVirt/ovir<wbr>t-engine-sdk-ruby/blob/master/<wbr>sdk/examples/asynchronous_inve<wbr>ntory.rb</a><br>
> <<a href="https://github.com/oVirt/ovirt-engine-sdk-ruby/blob/master/sdk/examples/asynchronous_inventory.rb" rel="noreferrer" target="_blank">https://github.com/oVirt/ovi<wbr>rt-engine-sdk-ruby/blob/master<wbr>/sdk/examples/asynchronous_inv<wbr>entory.rb</a>><br>
><br>
> So make sure that you take advantage of that in your clients. Sadly<br>
> pipe-lining is disabled by default in most browsers, so this isn't<br>
> helpful for JavaScript applications.<br>
><br>
><br>
> But we can try what we can do in moVirt about<br>
> this: <a href="https://github.com/oVirt/moVirt/issues/260" rel="noreferrer" target="_blank">https://github.com/oVirt/moVir<wbr>t/issues/260</a><br>
><br>
<br>
</div></div>Sure, I think there are plenty of asynchronous HTTP clients for Android,<br>
worth trying one of them. If you are brave enough you can even consider<br>
using the same library used in the Ruby SDK: libcurl. A bit of JNI here<br>
and there, and you are done.<br>
<span class="gmail-m_-971199036686354921m_4462347545091815787gmail-"><br>
><br>
><br>
> 4. HTTP/2 support.<br>
><br>
> The application server that we use, WildFly, supports HTTP/2, including<br>
> ALPN, out of the box, since version 10.1. We need a mechanism to<br>
> enable it:<br>
><br>
> core: Add support for enabling HTTP/2<br>
> <a href="https://gerrit.ovirt.org/74621" rel="noreferrer" target="_blank">https://gerrit.ovirt.org/7462<wbr>1</a><br>
><br>
> And then we need to get Apache out of the way, for API traffic, at<br>
> least. I think that is something we can do in the context of the engine<br>
> "podification" effort.<br>
><br>
> However, note that HTTP/2 won't have that big impact in performance for<br>
> applications that continue to use a synchronous/serial style of<br>
> interaction with the API.<br>
><br>
> On 03/24/2017 11:16 PM, Yaniv Kaul wrote:<br>
> ><br>
> ><br>
> > On Fri, Mar 24, 2017 at 8:57 PM, Martin Sivak <<a href="mailto:msivak@redhat.com" target="_blank">msivak@redhat.com</a> <mailto:<a href="mailto:msivak@redhat.com" target="_blank">msivak@redhat.com</a>><br>
</span><span class="gmail-m_-971199036686354921m_4462347545091815787gmail-">> > <mailto:<a href="mailto:msivak@redhat.com" target="_blank">msivak@redhat.com</a> <mailto:<a href="mailto:msivak@redhat.com" target="_blank">msivak@redhat.com</a>>>> wrote:<br>
> ><br>
> > > Current Apache used has only experimental module for it.<br>
> > > Undertow is supposed to have a better support. I wonder when/if we can drop<br>
> > > Apache...<br>
> ><br>
> > The last info I have about that from mperina is that we need Apache<br>
> > for kerberos support atm.<br>
> ><br>
> ><br>
> > I don't think we need it - I remember reading that Undertow does support<br>
> > it as well.<br>
> > The only issue is that there are probably 10 people in the world who<br>
> > know how to configure Undertow for Kerberos, while many do for Apache.<br>
> > And since we leave it for the user to configure...<br>
> > Y.<br>
> ><br>
> ><br>
> ><br>
> > Martin<br>
> ><br>
> > On Fri, Mar 24, 2017 at 5:30 PM, Yaniv Kaul <<a href="mailto:ykaul@redhat.com" target="_blank">ykaul@redhat.com</a> <mailto:<a href="mailto:ykaul@redhat.com" target="_blank">ykaul@redhat.com</a>><br>
</span><span class="gmail-m_-971199036686354921m_4462347545091815787gmail-">> > <mailto:<a href="mailto:ykaul@redhat.com" target="_blank">ykaul@redhat.com</a> <mailto:<a href="mailto:ykaul@redhat.com" target="_blank">ykaul@redhat.com</a>>>> wrote:<br>
> > ><br>
> > ><br>
> > > On Fri, Mar 24, 2017 at 6:43 PM, Martin Sivak <<a href="mailto:msivak@redhat.com" target="_blank">msivak@redhat.com</a> <mailto:<a href="mailto:msivak@redhat.com" target="_blank">msivak@redhat.com</a>><br>
</span><div><div class="gmail-m_-971199036686354921m_4462347545091815787gmail-h5">> > <mailto:<a href="mailto:msivak@redhat.com" target="_blank">msivak@redhat.com</a> <mailto:<a href="mailto:msivak@redhat.com" target="_blank">msivak@redhat.com</a>>>> wrote:<br>
> > >><br>
> > >> > 2: you can have more api gateways (e.g. more apis)<br>
> tailored for<br>
> > every<br>
> > >> > frontend. I don't think we need this - the current API serves<br>
> > us pretty<br>
> > >> > well<br>
> > >> > in every FE Im involved in. The only thing which I miss<br>
> is the data<br>
> > >> > aggregation.<br>
> > >><br>
> > >> So it does not serve us well. Aggregation of data is one<br>
> the usual<br>
> > >> points of using the gateway.<br>
> > >> Yes microservices are affected by this indeed, but so are<br>
> we because<br>
> > >> implementing the aggregation directly in the current engine<br>
> API layer<br>
> > >> is hard.<br>
> > >><br>
> > >> > So I would go back to the original topic of this thread - do<br>
> > some small<br>
> > >> > change which has a chance to be merged to the project and<br>
> helps<br>
> > us where<br>
> > >> > it<br>
> > >> > hurts.<br>
> > ><br>
> > ><br>
> > > I'm wondering if very specific additional REST API calls can<br>
> suffice.<br>
> > > For example, a 'Get VM + disks + NIC' API call seems<br>
> reasonable to<br>
> > add for<br>
> > > the various clients who commonly need it.<br>
> > ><br>
> > >><br>
> > >> Can a simple HTTP/2 to HTTP/AJP gateway be the simplest<br>
> solution? Our<br>
> > >> Apache might even have a module for it already.<br>
> > ><br>
> > ><br>
> > > Current Apache used has only experimental module for it.<br>
> > > Undertow is supposed to have a better support. I wonder<br>
> when/if we<br>
> > can drop<br>
> > > Apache...<br>
> > > Y.<br>
> > ><br>
> > >><br>
> > >> That way you can multiplex all the REST calls using a<br>
> single tcp<br>
> > >> connection (and a single SSL negotiation).<br>
> > >><br>
> > >> A custom SSO enabled service like that might be even better<br>
> as it<br>
> > >> would be able to skip the authentication<br>
> > >> layers too and that would lower the engine load. But I am<br>
> not sure it<br>
> > >> is possible with the current codebase.<br>
> > >><br>
> > >> Martin<br>
> > >><br>
> > >> On Fri, Mar 24, 2017 at 4:22 PM, Tomas Jelinek<br>
> > <<a href="mailto:tjelinek@redhat.com" target="_blank">tjelinek@redhat.com</a> <mailto:<a href="mailto:tjelinek@redhat.com" target="_blank">tjelinek@redhat.com</a>><br>
</div></div>> <mailto:<a href="mailto:tjelinek@redhat.com" target="_blank">tjelinek@redhat.com</a> <mailto:<a href="mailto:tjelinek@redhat.com" target="_blank">tjelinek@redhat.com</a>>>><br>
<span class="gmail-m_-971199036686354921m_4462347545091815787gmail-">> > >> wrote:<br>
> > >> ><br>
> > >> ><br>
> > >> > On Fri, Mar 24, 2017 at 3:58 PM, Martin Sivak<br>
> > <<a href="mailto:msivak@redhat.com" target="_blank">msivak@redhat.com</a> <mailto:<a href="mailto:msivak@redhat.com" target="_blank">msivak@redhat.com</a>><br>
</span><div><div class="gmail-m_-971199036686354921m_4462347545091815787gmail-h5">> <mailto:<a href="mailto:msivak@redhat.com" target="_blank">msivak@redhat.com</a> <mailto:<a href="mailto:msivak@redhat.com" target="_blank">msivak@redhat.com</a>>>> wrote:<br>
> > >> >><br>
> > >> >> > I feel like every REST API I've ever worked with has<br>
> had the<br>
> > >> >> > aggregation<br>
> > >> >> > +<br>
> > >> >> > projection problem. It's like we're trying to use REST<br>
> as a<br>
> > >> >> > replacement<br>
> > >> >> > for<br>
> > >> >> > SQL -- but the logic that executes the "SQL" lives in a<br>
> > browser now,<br>
> > >> >> > and<br>
> > >> >> > it<br>
> > >> >> > used to live on a server close to the DB. And REST isn't<br>
> > expressive<br>
> > >> >> > for<br>
> > >> >> > selecting data like SQL is.<br>
> > >> >><br>
> > >> >> The current industry solution I know about is called API<br>
> gateway..<br>
> > >> >> most of the big players have internal API with lots of low<br>
> > level stuff<br>
> > >> >> and then couple of external API gateways tailored to<br>
> what the<br>
> > client<br>
> > >> >> needs.<br>
> > >> >><br>
> > >> >> <a href="http://microservices.io/patterns/apigateway.html" rel="noreferrer" target="_blank">http://microservices.io/patter<wbr>ns/apigateway.html</a><br>
> <<a href="http://microservices.io/patterns/apigateway.html" rel="noreferrer" target="_blank">http://microservices.io/patt<wbr>erns/apigateway.html</a>><br>
> > <<a href="http://microservices.io/patterns/apigateway.html" rel="noreferrer" target="_blank">http://microservices.io/patt<wbr>erns/apigateway.html</a><br>
> <<a href="http://microservices.io/patterns/apigateway.html" rel="noreferrer" target="_blank">http://microservices.io/patt<wbr>erns/apigateway.html</a>>> (check the backend<br>
> > >> >> for frontend section)<br>
> > >> >><br>
> > >> >> This trend is also visible when you think about services<br>
> that<br>
> > offer<br>
> > >> >> API gateway management and billing like<br>
> > >> >> <a href="https://aws.amazon.com/api-gateway/" rel="noreferrer" target="_blank">https://aws.amazon.com/api-gat<wbr>eway/</a><br>
> <<a href="https://aws.amazon.com/api-gateway/" rel="noreferrer" target="_blank">https://aws.amazon.com/api-g<wbr>ateway/</a>><br>
> > <<a href="https://aws.amazon.com/api-gateway/" rel="noreferrer" target="_blank">https://aws.amazon.com/api-g<wbr>ateway/</a><br>
> <<a href="https://aws.amazon.com/api-gateway/" rel="noreferrer" target="_blank">https://aws.amazon.com/api-g<wbr>ateway/</a>>> or our very own<br>
> > >> >> <a href="https://www.3scale.net/" rel="noreferrer" target="_blank">https://www.3scale.net/</a><br>
> > >> ><br>
> > >> ><br>
> > >> > right, but the api gateway solves 2 problems:<br>
> > >> ><br>
> > >> > 1: if you have a microservice architecture it is hard for<br>
> > frontend to<br>
> > >> > talk<br>
> > >> > to 20 different moving services. So the gateway hides this<br>
> > complexity<br>
> > >> > behind<br>
> > >> > it. This is not the problem we have.<br>
> > >> ><br>
> > >> > 2: you can have more api gateways (e.g. more apis)<br>
> tailored for<br>
> > every<br>
> > >> > frontend. I don't think we need this - the current API serves<br>
> > us pretty<br>
> > >> > well<br>
> > >> > in every FE Im involved in. The only thing which I miss<br>
> is the data<br>
> > >> > aggregation.<br>
> > >> ><br>
> > >> > So I would go back to the original topic of this thread - do<br>
> > some small<br>
> > >> > change which has a chance to be merged to the project and<br>
> helps<br>
> > us where<br>
> > >> > it<br>
> > >> > hurts.<br>
> > >> ><br>
> > >> >><br>
> > >> >><br>
> > >> >><br>
> > >> >> Martin<br>
> > >> >><br>
> > >> >> On Fri, Mar 24, 2017 at 3:47 PM, Greg Sheremeta<br>
> > <<a href="mailto:gshereme@redhat.com" target="_blank">gshereme@redhat.com</a> <mailto:<a href="mailto:gshereme@redhat.com" target="_blank">gshereme@redhat.com</a>><br>
</div></div>> <mailto:<a href="mailto:gshereme@redhat.com" target="_blank">gshereme@redhat.com</a> <mailto:<a href="mailto:gshereme@redhat.com" target="_blank">gshereme@redhat.com</a>>>><br>
<span class="gmail-m_-971199036686354921m_4462347545091815787gmail-">> > >> >> wrote:<br>
> > >> >> > I feel like every REST API I've ever worked with has had the<br>
> > >> >> > aggregation<br>
> > >> >> > +<br>
> > >> >> > projection problem. It's like we're trying to use REST as a<br>
> > >> >> > replacement<br>
> > >> >> > for<br>
> > >> >> > SQL -- but the logic that executes the "SQL" lives in a<br>
> > browser now,<br>
> > >> >> > and<br>
> > >> >> > it<br>
> > >> >> > used to live on a server close to the DB. And REST isn't<br>
> > expressive<br>
> > >> >> > for<br>
> > >> >> > selecting data like SQL is.<br>
> > >> >> ><br>
> > >> >> > There must be some industry solution to this "I want to do<br>
> > SQL over<br>
> > >> >> > REST"<br>
> > >> >> > problem.<br>
> > >> >> ><br>
> > >> >> > On Fri, Mar 24, 2017 at 5:54 AM, Martin Sivak<br>
> > <<a href="mailto:msivak@redhat.com" target="_blank">msivak@redhat.com</a> <mailto:<a href="mailto:msivak@redhat.com" target="_blank">msivak@redhat.com</a>><br>
</span>> <mailto:<a href="mailto:msivak@redhat.com" target="_blank">msivak@redhat.com</a> <mailto:<a href="mailto:msivak@redhat.com" target="_blank">msivak@redhat.com</a>>>><br>
<div><div class="gmail-m_-971199036686354921m_4462347545091815787gmail-h5">> > >> >> > wrote:<br>
> > >> >> >><br>
> > >> >> >> > for quite some time I have been more or less<br>
> involved in<br>
> > >> >> >> > development<br>
> > >> >> >> > of<br>
> > >> >> >> > various UIs for oVirt based entirely on the oVirt's<br>
> REST API<br>
> > >> >> >> > ranging<br>
> > >> >> >> > from<br>
> > >> >> >> > the quite mature moVirt [1] through some cockpit<br>
> > extensions to a<br>
> > >> >> >> > young<br>
> > >> >> >> > and<br>
> > >> >> >> > experimental user portal replacement [2].<br>
> > >> >> >><br>
> > >> >> >> oVirt optimizer has the same issue..<br>
> > >> >> >><br>
> > >> >> >> > 2: add some tiny service which would just accept a<br>
> list of<br>
> > >> >> >> > queries,<br>
> > >> >> >> > execute<br>
> > >> >> >> > them locally (but using real HTTP requests) and<br>
> return in one<br>
> > >> >> >> > bulk. A<br>
> > >> >> >> > naive<br>
> > >> >> >> > implementation just to give a sense of what I mean of<br>
> > this would<br>
> > >> >> >> > be a<br>
> > >> >> >> > shell<br>
> > >> >> >> > script getting list of strings like<br>
> > >> >> >> ><br>
> "<a href="https://localhost/ovirt-engine/api/vms/123/sessions" rel="noreferrer" target="_blank">https://localhost/ovirt-engi<wbr>ne/api/vms/123/sessions</a><br>
> <<a href="https://localhost/ovirt-engine/api/vms/123/sessions" rel="noreferrer" target="_blank">https://localhost/ovirt-engi<wbr>ne/api/vms/123/sessions</a>><br>
> > <<a href="https://localhost/ovirt-engine/api/vms/123/sessions" rel="noreferrer" target="_blank">https://localhost/ovirt-engi<wbr>ne/api/vms/123/sessions</a><br>
> <<a href="https://localhost/ovirt-engine/api/vms/123/sessions" rel="noreferrer" target="_blank">https://localhost/ovirt-engi<wbr>ne/api/vms/123/sessions</a>>>" iterate over<br>
> > >> >> >> > them<br>
> > >> >> >> > and<br>
> > >> >> >> > do a curl request for each, mangle the results into one<br>
> > string and<br>
> > >> >> >> > return<br>
> > >> >> >> > (credits for this idea to msivak). Easy to implement,<br>
> > possibility<br>
> > >> >> >> > to<br>
> > >> >> >> > add<br>
> > >> >> >> > also projections later to save some bandwidth. But the<br>
> > API would<br>
> > >> >> >> > anyway<br>
> > >> >> >> > be<br>
> > >> >> >> > hammered by bunch of queries, only the network<br>
> roundtrip<br>
> > would be<br>
> > >> >> >> > saved.<br>
> > >> >> >><br>
> > >> >> >> The biggest cost for (especially mobile) clients is<br>
> the cost of<br>
> > >> >> >> establishing new SSL connection. SSL is also pretty<br>
> > expensive on the<br>
> > >> >> >> server side.<br>
> > >> >> >><br>
> > >> >> >> So running the aggregation service on the<br>
> ovirt-engine machine<br>
> > >> >> >> (behind<br>
> > >> >> >> Apache) means the client will do a single SSL request<br>
> with<br>
> > list of N<br>
> > >> >> >> urls and the local "reverse-proxy" will perform single<br>
> > >> >> >> authentication<br>
> > >> >> >> and N plain HTTP requests (or even better - AJP). It<br>
> won't<br>
> > remove<br>
> > >> >> >> any<br>
> > >> >> >> time from the actual command run time, but it will reduce<br>
> > protocol<br>
> > >> >> >> overhead.<br>
> > >> >> >><br>
> > >> >> >> I think this is the simplest first step that requires<br>
> almost no<br>
> > >> >> >> change<br>
> > >> >> >> to existing infrastructure.<br>
> > >> >> >><br>
> > >> >> >> --<br>
> > >> >> >> Martin Sivak<br>
> > >> >> >> SLA / oVirt<br>
> > >> >> >><br>
> > >> >> >> On Fri, Mar 24, 2017 at 10:20 AM, Tomas Jelinek<br>
> > >> >> >> <<a href="mailto:tjelinek@redhat.com" target="_blank">tjelinek@redhat.com</a> <mailto:<a href="mailto:tjelinek@redhat.com" target="_blank">tjelinek@redhat.com</a>><br>
</div></div>> <mailto:<a href="mailto:tjelinek@redhat.com" target="_blank">tjelinek@redhat.com</a> <mailto:<a href="mailto:tjelinek@redhat.com" target="_blank">tjelinek@redhat.com</a>>>><br>
<div><div class="gmail-m_-971199036686354921m_4462347545091815787gmail-h5">> > >> >> >> wrote:<br>
> > >> >> >> > Hi All,<br>
> > >> >> >> ><br>
> > >> >> >> > for quite some time I have been more or less<br>
> involved in<br>
> > >> >> >> > development<br>
> > >> >> >> > of<br>
> > >> >> >> > various UIs for oVirt based entirely on the oVirt's<br>
> REST API<br>
> > >> >> >> > ranging<br>
> > >> >> >> > from<br>
> > >> >> >> > the quite mature moVirt [1] through some cockpit<br>
> > extensions to a<br>
> > >> >> >> > young<br>
> > >> >> >> > and<br>
> > >> >> >> > experimental user portal replacement [2].<br>
> > >> >> >> ><br>
> > >> >> >> > One issue we hit over and over again is the missing<br>
> data<br>
> > >> >> >> > aggregation.<br>
> > >> >> >> > In<br>
> > >> >> >> > the<br>
> > >> >> >> > 3.x era we used to use in moVirt the detail=something<br>
> > >> >> >> > api to get the disks and nics of the VM, something<br>
> like:<br>
> > >> >> >> ><br>
> > >> >> >> > GET /ovirt-engine/api/vms<br>
> > >> >> >> > Accept: application/json; detail=disks<br>
> > >> >> >> ><br>
> > >> >> >> > This allowed us to store this data in local database<br>
> > leading to<br>
> > >> >> >> > great<br>
> > >> >> >> > user<br>
> > >> >> >> > experience. Since this feature has been removed in 4.x<br>
> > API [3]<br>
> > >> >> >> > we needed to retire to a different solution. When<br>
> the VM<br>
> > detail is<br>
> > >> >> >> > selected<br>
> > >> >> >> > by the user, start loading the disks and nics and hope<br>
> > the user<br>
> > >> >> >> > will not be fast enough to see the delay. The UX is<br>
> > slightly worse<br>
> > >> >> >> > bug<br>
> > >> >> >> > kinda<br>
> > >> >> >> > acceptable.<br>
> > >> >> >> ><br>
> > >> >> >> > We hit this issue harder in the new user portal [2],<br>
> > because we<br>
> > >> >> >> > already<br>
> > >> >> >> > have<br>
> > >> >> >> > the VM cached and show the whole VM in one screen.<br>
> So, if<br>
> > you pick<br>
> > >> >> >> > it,<br>
> > >> >> >> > you<br>
> > >> >> >> > will get it's details immediately.<br>
> > >> >> >> > But, since you don't have all the details, we need<br>
> to do an<br>
> > >> >> >> > additional<br>
> > >> >> >> > call<br>
> > >> >> >> > (two actually) to load this data and they start to<br>
> appear<br>
> > later.<br>
> > >> >> >> > So, something which would be very fast and smooth<br>
> starts<br>
> > to feel<br>
> > >> >> >> > sluggish.<br>
> > >> >> >> ><br>
> > >> >> >> > Recently, we hit this issue again which forced us to<br>
> > sacrifice the<br>
> > >> >> >> > UX<br>
> > >> >> >> > even<br>
> > >> >> >> > more - it is the "console in use" feature of user<br>
> portal.<br>
> > >> >> >> > The use case is this:<br>
> > >> >> >> > - if the console is already taken by some user,<br>
> there are<br>
> > >> >> >> > complications<br>
> > >> >> >> > if<br>
> > >> >> >> > other current user tryes to take it as well (will avoid<br>
> > details<br>
> > >> >> >> > about<br>
> > >> >> >> > settings and permissins involved, but long story short,<br>
> > the user<br>
> > >> >> >> > will<br>
> > >> >> >> > probably not be allowed to connect to it. The<br>
> "probably"<br>
> > is the<br>
> > >> >> >> > key<br>
> > >> >> >> > here<br>
> > >> >> >> > since we can not do any intelligent decision in<br>
> advance,<br>
> > we can<br>
> > >> >> >> > only<br>
> > >> >> >> > warn<br>
> > >> >> >> > the user that the console is taken).<br>
> > >> >> >> > - in the current GWT user portal, if the VM's<br>
> console is<br>
> > taken, it<br>
> > >> >> >> > is<br>
> > >> >> >> > shown<br>
> > >> >> >> > on the VM's "box" that "console is taken". This was<br>
> a highly<br>
> > >> >> >> > requested<br>
> > >> >> >> > feature<br>
> > >> >> >> > - to get this information using the current REST<br>
> API, we<br>
> > need to<br>
> > >> >> >> > go<br>
> > >> >> >> > to<br>
> > >> >> >> > the<br>
> > >> >> >> > /vms/<vmid>/sessions subcollection. To get this for all<br>
> > VMs, it<br>
> > >> >> >> > would<br>
> > >> >> >> > be<br>
> > >> >> >> > doing N queries per poll which we can not afford<br>
> > >> >> >> > - so the current PR [4] will probably end up to only<br>
> > check it on<br>
> > >> >> >> > the<br>
> > >> >> >> > attempt<br>
> > >> >> >> > to connect to the console warning the user. Maybe<br>
> it will<br>
> > be also<br>
> > >> >> >> > shown<br>
> > >> >> >> > in<br>
> > >> >> >> > Vm details. But the UX in case the user will look for a<br>
> > VM which<br>
> > >> >> >> > has<br>
> > >> >> >> > free<br>
> > >> >> >> > console will suffer significantly (e.g. try one by one<br>
> > until some<br>
> > >> >> >> > opens<br>
> > >> >> >> > or<br>
> > >> >> >> > look at details one by one to see if the warning<br>
> appears<br>
> > (with a<br>
> > >> >> >> > delay))<br>
> > >> >> >> ><br>
> > >> >> >> > I understand that embedding the details of the VM<br>
> to the<br>
> > response<br>
> > >> >> >> > comes<br>
> > >> >> >> > with<br>
> > >> >> >> > a cost, namely:<br>
> > >> >> >> > - performance hit<br>
> > >> >> >> > - complexity of the API code<br>
> > >> >> >> > - the "cleanness" of REST suffers<br>
> > >> >> >> ><br>
> > >> >> >> > But I think we should seriously consider to provide<br>
> some<br>
> > option to<br>
> > >> >> >> > data<br>
> > >> >> >> > aggregation.<br>
> > >> >> >> ><br>
> > >> >> >> > I know this has been discussed many times with no<br>
> result,<br>
> > but I<br>
> > >> >> >> > think<br>
> > >> >> >> > it<br>
> > >> >> >> > is<br>
> > >> >> >> > time to bring this topic up again. I'll try to<br>
> summarize the<br>
> > >> >> >> > (failed)<br>
> > >> >> >> > attempts tried so far:<br>
> > >> >> >> > - the detail=<something> parameter with ad-hoc<br>
> embedding<br>
> > of data.<br>
> > >> >> >> > This<br>
> > >> >> >> > has<br>
> > >> >> >> > been there and removed in 4.0 [3]<br>
> > >> >> >> > - the DoctorREST project - e.g. a proxy above the<br>
> current<br>
> > api. The<br>
> > >> >> >> > idea<br>
> > >> >> >> > was<br>
> > >> >> >> > to create a service which will be independent of<br>
> the engine<br>
> > >> >> >> > itself,<br>
> > >> >> >> > will<br>
> > >> >> >> > locally poll the engine's REST, store all data in local<br>
> > (mongo)DB<br>
> > >> >> >> > and<br>
> > >> >> >> > provide a rich api with aggregations and<br>
> projections and push<br>
> > >> >> >> > notifications.<br>
> > >> >> >> > This polling of everything to get the data to<br>
> DoctorREST<br>
> > proved to<br>
> > >> >> >> > be<br>
> > >> >> >> > pretty<br>
> > >> >> >> > costy, so also a more invasive approach of pushing<br>
> data from<br>
> > >> >> >> > engine<br>
> > >> >> >> > to<br>
> > >> >> >> > doctor has been discused [5]. None of this two<br>
> approaches<br>
> > have<br>
> > >> >> >> > been<br>
> > >> >> >> > accepted<br>
> > >> >> >> > (too complicated, too invasive).<br>
> > >> >> >> > - writing some custom ad-hoc servlet serving only a<br>
> > purpose of one<br>
> > >> >> >> > frontend<br>
> > >> >> >> > - this is actually there for the dashboard, but it<br>
> is not a<br>
> > >> >> >> > generic<br>
> > >> >> >> > solution<br>
> > >> >> >> > for the other frontends and we really should not<br>
> develop<br>
> > custom<br>
> > >> >> >> > "APIs"<br>
> > >> >> >> > for<br>
> > >> >> >> > every frontend<br>
> > >> >> >> > - there were some other proposals discussed (some<br>
> 3th party<br>
> > >> >> >> > solutions<br>
> > >> >> >> > etc)<br>
> > >> >> >> > but I think none of them made it even to a PoC<br>
> > >> >> >> ><br>
> > >> >> >> > So, now I would try again and try small to get at<br>
> least some<br>
> > >> >> >> > benefit.<br>
> > >> >> >> > I<br>
> > >> >> >> > see<br>
> > >> >> >> > 2 paths we could try:<br>
> > >> >> >> > 1: embed something which burns us immediatly, e.g. the<br>
> > /sessions<br>
> > >> >> >> > into<br>
> > >> >> >> > VMs. I<br>
> > >> >> >> > really liked the ;detail=sessions approach, could<br>
> we move<br>
> > it back?<br>
> > >> >> >> > 2: add some tiny service which would just accept a<br>
> list of<br>
> > >> >> >> > queries,<br>
> > >> >> >> > execute<br>
> > >> >> >> > them locally (but using real HTTP requests) and<br>
> return in one<br>
> > >> >> >> > bulk. A<br>
> > >> >> >> > naive<br>
> > >> >> >> > implementation just to give a sense of what I mean of<br>
> > this would<br>
> > >> >> >> > be a<br>
> > >> >> >> > shell<br>
> > >> >> >> > script getting list of strings like<br>
> > >> >> >> ><br>
> "<a href="https://localhost/ovirt-engine/api/vms/123/sessions" rel="noreferrer" target="_blank">https://localhost/ovirt-engi<wbr>ne/api/vms/123/sessions</a><br>
> <<a href="https://localhost/ovirt-engine/api/vms/123/sessions" rel="noreferrer" target="_blank">https://localhost/ovirt-engi<wbr>ne/api/vms/123/sessions</a>><br>
> > <<a href="https://localhost/ovirt-engine/api/vms/123/sessions" rel="noreferrer" target="_blank">https://localhost/ovirt-engi<wbr>ne/api/vms/123/sessions</a><br>
> <<a href="https://localhost/ovirt-engine/api/vms/123/sessions" rel="noreferrer" target="_blank">https://localhost/ovirt-engi<wbr>ne/api/vms/123/sessions</a>>>" iterate over<br>
> > >> >> >> > them<br>
> > >> >> >> > and<br>
> > >> >> >> > do a curl request for each, mangle the results into one<br>
> > string and<br>
> > >> >> >> > return<br>
> > >> >> >> > (credits for this idea to msivak). Easy to implement,<br>
> > possibility<br>
> > >> >> >> > to<br>
> > >> >> >> > add<br>
> > >> >> >> > also projections later to save some bandwidth. But the<br>
> > API would<br>
> > >> >> >> > anyway<br>
> > >> >> >> > be<br>
> > >> >> >> > hammered by bunch of queries, only the network<br>
> roundtrip<br>
> > would be<br>
> > >> >> >> > saved.<br>
> > >> >> >> > 3: any other simple approaches?<br>
> > >> >> >> ><br>
> > >> >> >> > I honestly prefer the first approach. It is not<br>
> > beautiful, it is<br>
> > >> >> >> > not<br>
> > >> >> >> > REST-ful, but it is easy to implement, very<br>
> pragmatic and<br>
> > useful.<br>
> > >> >> >> > What do you think?<br>
> > >> >> >> ><br>
> > >> >> >> > Thank you and sorry for the long mail :)<br>
> > >> >> >> > Tomas<br>
> > >> >> >> ><br>
> > >> >> >> > [1]: <a href="https://github.com/oVirt/moVirt" rel="noreferrer" target="_blank">https://github.com/oVirt/moVir<wbr>t</a><br>
> <<a href="https://github.com/oVirt/moVirt" rel="noreferrer" target="_blank">https://github.com/oVirt/moV<wbr>irt</a>><br>
> > <<a href="https://github.com/oVirt/moVirt" rel="noreferrer" target="_blank">https://github.com/oVirt/moV<wbr>irt</a><br>
> <<a href="https://github.com/oVirt/moVirt" rel="noreferrer" target="_blank">https://github.com/oVirt/moV<wbr>irt</a>>><br>
> > >> >> >> > [2]: <a href="https://github.com/oVirt/ovirt-web-ui" rel="noreferrer" target="_blank">https://github.com/oVirt/ovirt<wbr>-web-ui</a><br>
> <<a href="https://github.com/oVirt/ovirt-web-ui" rel="noreferrer" target="_blank">https://github.com/oVirt/ovi<wbr>rt-web-ui</a>><br>
> > <<a href="https://github.com/oVirt/ovirt-web-ui" rel="noreferrer" target="_blank">https://github.com/oVirt/ovi<wbr>rt-web-ui</a><br>
> <<a href="https://github.com/oVirt/ovirt-web-ui" rel="noreferrer" target="_blank">https://github.com/oVirt/ovi<wbr>rt-web-ui</a>>><br>
> > >> >> >> > [3]: <a href="https://gerrit.ovirt.org/#/c/61260" rel="noreferrer" target="_blank">https://gerrit.ovirt.org/#/c/6<wbr>1260</a><br>
> <<a href="https://gerrit.ovirt.org/#/c/61260" rel="noreferrer" target="_blank">https://gerrit.ovirt.org/#/c<wbr>/61260</a>><br>
> > <<a href="https://gerrit.ovirt.org/#/c/61260" rel="noreferrer" target="_blank">https://gerrit.ovirt.org/#/c<wbr>/61260</a><br>
> <<a href="https://gerrit.ovirt.org/#/c/61260" rel="noreferrer" target="_blank">https://gerrit.ovirt.org/#/c<wbr>/61260</a>>><br>
> > >> >> >> > [4]:<br>
> <a href="https://github.com/oVirt/ovirt-web-ui/pull/106/" rel="noreferrer" target="_blank">https://github.com/oVirt/ovir<wbr>t-web-ui/pull/106/</a><br>
> <<a href="https://github.com/oVirt/ovirt-web-ui/pull/106/" rel="noreferrer" target="_blank">https://github.com/oVirt/ovi<wbr>rt-web-ui/pull/106/</a>><br>
> > <<a href="https://github.com/oVirt/ovirt-web-ui/pull/106/" rel="noreferrer" target="_blank">https://github.com/oVirt/ovi<wbr>rt-web-ui/pull/106/</a><br>
> <<a href="https://github.com/oVirt/ovirt-web-ui/pull/106/" rel="noreferrer" target="_blank">https://github.com/oVirt/ovi<wbr>rt-web-ui/pull/106/</a>>><br>
> > >> >> >> > [5]: <a href="https://gerrit.ovirt.org/#/c/45233/" rel="noreferrer" target="_blank">https://gerrit.ovirt.org/#/c/4<wbr>5233/</a><br>
> <<a href="https://gerrit.ovirt.org/#/c/45233/" rel="noreferrer" target="_blank">https://gerrit.ovirt.org/#/c<wbr>/45233/</a>><br>
> > <<a href="https://gerrit.ovirt.org/#/c/45233/" rel="noreferrer" target="_blank">https://gerrit.ovirt.org/#/c<wbr>/45233/</a><br>
> <<a href="https://gerrit.ovirt.org/#/c/45233/" rel="noreferrer" target="_blank">https://gerrit.ovirt.org/#/c<wbr>/45233/</a>>><br>
> > >> >> >> ><br>
> > >> >> >> ><br>
> > >> >> >> > ______________________________<wbr>_________________<br>
> > >> >> >> > Devel mailing list<br>
> > >> >> >> > <a href="mailto:Devel@ovirt.org" target="_blank">Devel@ovirt.org</a> <mailto:<a href="mailto:Devel@ovirt.org" target="_blank">Devel@ovirt.org</a>><br>
</div></div>> <mailto:<a href="mailto:Devel@ovirt.org" target="_blank">Devel@ovirt.org</a> <mailto:<a href="mailto:Devel@ovirt.org" target="_blank">Devel@ovirt.org</a>>><br>
<span class="gmail-m_-971199036686354921m_4462347545091815787gmail-">> > >> >> >> > <a href="http://lists.ovirt.org/mailman/listinfo/devel" rel="noreferrer" target="_blank">http://lists.ovirt.org/mailman<wbr>/listinfo/devel</a><br>
> <<a href="http://lists.ovirt.org/mailman/listinfo/devel" rel="noreferrer" target="_blank">http://lists.ovirt.org/mailm<wbr>an/listinfo/devel</a>><br>
> > <<a href="http://lists.ovirt.org/mailman/listinfo/devel" rel="noreferrer" target="_blank">http://lists.ovirt.org/mailm<wbr>an/listinfo/devel</a><br>
> <<a href="http://lists.ovirt.org/mailman/listinfo/devel" rel="noreferrer" target="_blank">http://lists.ovirt.org/mailm<wbr>an/listinfo/devel</a>>><br>
> > >> >> >> ______________________________<wbr>_________________<br>
> > >> >> >> Devel mailing list<br>
> > >> >> >> <a href="mailto:Devel@ovirt.org" target="_blank">Devel@ovirt.org</a> <mailto:<a href="mailto:Devel@ovirt.org" target="_blank">Devel@ovirt.org</a>><br>
</span>> <mailto:<a href="mailto:Devel@ovirt.org" target="_blank">Devel@ovirt.org</a> <mailto:<a href="mailto:Devel@ovirt.org" target="_blank">Devel@ovirt.org</a>>><br>
> > >> >> >> <a href="http://lists.ovirt.org/mailman/listinfo/devel" rel="noreferrer" target="_blank">http://lists.ovirt.org/mailman<wbr>/listinfo/devel</a><br>
> <<a href="http://lists.ovirt.org/mailman/listinfo/devel" rel="noreferrer" target="_blank">http://lists.ovirt.org/mailm<wbr>an/listinfo/devel</a>><br>
<span class="gmail-m_-971199036686354921m_4462347545091815787gmail-">> > <<a href="http://lists.ovirt.org/mailman/listinfo/devel" rel="noreferrer" target="_blank">http://lists.ovirt.org/mailm<wbr>an/listinfo/devel</a><br>
> <<a href="http://lists.ovirt.org/mailman/listinfo/devel" rel="noreferrer" target="_blank">http://lists.ovirt.org/mailm<wbr>an/listinfo/devel</a>>><br>
> > >> >> ><br>
> > >> >> ><br>
> > >> >> ><br>
> > >> >> ><br>
> > >> >> > --<br>
> > >> >> > Greg Sheremeta, MBA<br>
> > >> >> > Red Hat, Inc.<br>
> > >> >> > Sr. Software Engineer<br>
> > >> >> > <a href="mailto:gshereme@redhat.com" target="_blank">gshereme@redhat.com</a> <mailto:<a href="mailto:gshereme@redhat.com" target="_blank">gshereme@redhat.com</a>><br>
</span>> <mailto:<a href="mailto:gshereme@redhat.com" target="_blank">gshereme@redhat.com</a> <mailto:<a href="mailto:gshereme@redhat.com" target="_blank">gshereme@redhat.com</a>>><br>
<span class="gmail-m_-971199036686354921m_4462347545091815787gmail-">> > >> ><br>
> > >> ><br>
> > >> ______________________________<wbr>_________________<br>
> > >> Devel mailing list<br>
> > >> <a href="mailto:Devel@ovirt.org" target="_blank">Devel@ovirt.org</a> <mailto:<a href="mailto:Devel@ovirt.org" target="_blank">Devel@ovirt.org</a>><br>
</span>> <mailto:<a href="mailto:Devel@ovirt.org" target="_blank">Devel@ovirt.org</a> <mailto:<a href="mailto:Devel@ovirt.org" target="_blank">Devel@ovirt.org</a>>><br>
> > >> <a href="http://lists.ovirt.org/mailman/listinfo/devel" rel="noreferrer" target="_blank">http://lists.ovirt.org/mailman<wbr>/listinfo/devel</a><br>
> <<a href="http://lists.ovirt.org/mailman/listinfo/devel" rel="noreferrer" target="_blank">http://lists.ovirt.org/mailm<wbr>an/listinfo/devel</a>><br>
> > <<a href="http://lists.ovirt.org/mailman/listinfo/devel" rel="noreferrer" target="_blank">http://lists.ovirt.org/mailm<wbr>an/listinfo/devel</a><br>
> <<a href="http://lists.ovirt.org/mailman/listinfo/devel" rel="noreferrer" target="_blank">http://lists.ovirt.org/mailm<wbr>an/listinfo/devel</a>>><br>
> > ><br>
> > ><br>
> ><br>
> ><br>
><br>
><br>
<br>
</blockquote></div></div></div><br></div></div>
<br>______________________________<wbr>_________________<br>
Devel mailing list<br>
<a href="mailto:Devel@ovirt.org" target="_blank">Devel@ovirt.org</a><br>
<a href="http://lists.ovirt.org/mailman/listinfo/devel" rel="noreferrer" target="_blank">http://lists.ovirt.org/mailman<wbr>/listinfo/devel</a><br></blockquote></div><br><br clear="all"><div><br></div>-- <br><div class="gmail-m_-971199036686354921gmail_signature"><div dir="ltr"><div>Greg Sheremeta, MBA<br>Red Hat, Inc.<br>Sr. Software Engineer<br><a href="mailto:gshereme@redhat.com" target="_blank">gshereme@redhat.com</a><br></div></div></div>
</div></div></div>