Google Anthos vs PCF vs Openshift: A Roundtable Discussion

you welcome guys till I am Tim D'Alessio I am the CTO over here at ecolution and I run our cloud practice so I got a very special treat for you guys today I put together a little panel of some of my experts over at ecolution and they will represent some different competing technologies so with anthos now in the marketplace there are some competing technologies to how it works and some of the advantages there so I thought it'd be good to put a little panel together here so we can have your discussion so let's introduce you guys so once you start with Muys give me a little bit background and what you do at ecolution hi i'm moist Khan I work at occlusion in Singapore as a cloud engineer my main area of expertise for the last five years has been pivotal cloud foundry and a Jing and operating and everything and for the last few months I've been working on gke and great over you pretty sure and I was working as a DevOps engineer in my last eight months in Cappadocia and previous before that I was working with open ship platform and managing the platform and maintain the platform so my main experience these companies products cool and the man with the epic beard I'm Brian frost I'm in Oklahoma City in the US also with technician who strangely enough and and I've been working with anthos now for a couple months done a done a fair fair bit of installs just just did another this week so been been working with it quite a bit lately all right great so let's kick it off with a question so it shouldn't be too hard it should be all within your wheelhouse here so let's see what you got so no kubernetes is leading the industry in providing an environment that supports scheduling and running docker containers and give existing kubernetes tools and applications be easily integrated within the platform that you are representing so let's start over with mini Louie's on the PC PCF side so PCF in its current state supports docker or but it in a very limited fashion can push docker images as applications but they're still you know they must still adhere to the constraints and rules that apply to regular conforme applications but it's in no way a kubernetes compatible or based on kubernetes so it's it's not going to work with you know traditional kubernetes tools like hell we will have a separate product for a more kubernetes focused approach which is called PKS and sort of it works alongside pcs sounds good so what about on the open ship side yeah also build on top of doctor and poverty's and the main thing about open ship is that you have to use rel VM or same toys as your mother notes so that means you cannot use other Linux distros going to copper ship and the other thing I see is if you are going to in the stock I mean investigate on poker chip it has fixed coupon security policies so that means you can't simply get an image from docker hub and run there because participate default it's not allowed to run high school there are you know for initial work that you need to adjust you need to understand before jump it with your function and then finally mr. an throws man well well the beauty of anthos is really that it and particularly we're talking about the gke on-prem portion of anthos and that that it's a a curator Nettie's environment by the people who created kubernetes and so it natively supports you know the wealth of Industry and the open source communities available tools and applications so everything's built to to standardize the cloud native foundations kubernetes standards so it really is it's as kubernetes as you can get so it'll run everything that that is available to for standard kubernetes all right thanks for the answers so let's been one of the next question then so is the management layer of your platform capable of managing both cloud and on premise data centers early alpha on the PCF side of PCF instance is very specific to your cloud environment or to your data center if you're going on Prem there's no management layer that spreads across multiple PCF instances each is very isolated very independent but regarding the hybrid approach PCF newer versions of PCF do have support for hto which is a service mesh to bridge applications that are running across different cloud environments or between on-prem and the cloud just as a follow-up so what about managing in one place would it what if we have multiple environments is there any place or centralized location that you could actually view all these different foundations coming Britain to club PCF there's no such thing like there's no tool provided by pivotal for the centralized management of multiple PCF instances whether it's on Prem or on the cloud or anywhere ok got it on the open ship side what about the managing on the cloud and the erm what does that look like yeah open container part I mean it has several flavors so if you talk about container platform you can run your web the nodes conferences as well as cloud and you can deploy clusters in anywhere any cloud provider open container platform ocean and I have seen a demo that you can route traffic using a five is a five support some of the basic features so for example you deploy your three clusters in one premises or AWS on TCP and using f5 router you can route traffic to those three clusters management layer one management layer I have I have seen I have heard that this is a power management tool to manage different platforms but I haven't have experience on that so I haven't seen it by myself but I have read through that there's a cloud management tool for that but I haven't seen it okay and then finally on the anthro side what does that look like would answer directly your question it's a little bit tricky because the ant those part of things you is the the on premise portion of GCP and gke so what what you do end up with is you do end up with GCP and its management interface for kubernetes actually applies to both your on-prem and your cloud gke instances you get that you get that single pane of glass where you all your clusters they show up together they're all all in order you go look at your different entities and they're they're either mixed together if you want to see them that way or you see all those entities separately and and already the portions of Ischia that are built into two gke on-prem allow you to move things around and have like a global name spacing but some of the tools are still coming out it looks like they're scheduled for next quarter to allow you to move move into these between workloads both in the cloud and all the same length well I say fairly seamlessly we're those early stages of development on pieces of this but but the things available to you already really enable that that hybrid hybrid cloud environment so you can you can manage between clusters both both once hosted in the cloud and hosted on-site from one place it's great see I told you these guys these questions would be easy no problem so next question let's talk about cost and you know expenses so you know how expensive are these implementations to run I mean is this something that I'm gonna be breaking the bank on or are there are these affordable those let's kick it off with PCF pricing on PCF is based on the number of application instances you plan to run so you can have regardless of the infrastructure that if provision for PCF so if you're running 100 applications of thousand applications that's what the cost is based off and but in addition to you like paying for the application instances you also get access to add-on services from pivotal like you know database services like my sequel and messaging services and other things that allow you to integrate with external services and if you're looking for something very specific people to also has a lot of tiles available but those are licensed separately from the main one is that include your lower environments - you have to pay for that or is that a-- included any environment any number of application instances that you run are factored into your cost okay interesting and how is the pricing structure and open shipped so like I said different flavors so one is quite of actual overtly for my actual so that one will cost roughly around forty three K per year I'm talking about four nodes of four course and 16gb of RAM so it will cost roughly around 43 km and there's another quick product which is dedicated chick dedicated on a power provider for semolina bees so it poses two flavours here that is single and high will be highly available clusters so that will cost you around 36 K for a single available cluster and 81k here high available custom okay other product that you can run on anywhere you like on your on-premises or one your cloud provider and that is what you container platform previously pose soft and face of subscription I heard that they are moving to Kobe subscription but I don't know why's the situation there and pricing Mosel not I I don't have the direct pricing of at the moment I'm previously posed around 10k per to suffered when I was doing open ship but I think that have been changed now I see so after keeper keep our ears open for that okay what about anthos how does that compare in the pricing model so they do price there's a little bit differently they price there's based on V CPUs since it is a V center environment the the virtual CPUs that you have defined here or nodes orally where the cost comes from and you you ask a question I think to moiez about if you're lower environments are priced the same as your your production environments and so they they are priced the same but the fact that they're based on V CPUs dev and and just your general non prod environments aren't going to be heavy cpu utilizers so so it gives you a real break on the lower environments and and how what they actually cost you but it looks like in well the pricing still really in flux so so in some of the early meetings they were discussed to be around 10k a month for a hundred V CPUs that you have defined okay so I I can't really commit to that but I can commit to having heard that it's being one of the pricing models that it's being discussed I mean I know anthos is still pretty new so I'm sure they're trying to flush out a lot of that great so let's move on to another big aspect of you know app as environment and that's really important especially to your operators you know what does the monitoring look like are there any in monitoring solutions that come out of the box or are these things that we're gonna have to bolt on in the future well let's start over with PCF so what does the monitoring look like in that environment yeah so PCF on its own and doesn't have any strong monitoring tools Britain but people does provide tools like HealthWatch and PCF metrics as add-ons to your PCF deployment as part of your pivotal license in addition to that cloud foundry in general can integrate with you know popular monitoring tools yeah okay good what about open about 3.2 so at that moment they have Oklahoma trains and Cassandra to use auto scaling so that that's a problem we had at that time there were good monitoring solution for Pacific platform I'm talking about version 3.2 but now they have introduced Prometheus and Ravana by default it is pre-configured in the new versions so now they are rely on for ETS and Ravana motoring stack but pretty early days there was not much motoring related stuff indeed great sounds good finally entos what is what does that look like in the monitor space but first off that's pretty cool that openshift is doing that because that's that's the same same tool set which I really is they're the leaders in the market Prometheus and ember fauna they're built in day one and so so they're they're always watching your metrics for you I'm always really partial to a graph on how quickly you can get up and running with with metrics and and you already have pre-built dashboards that show you the things that are important to kubernetes and then Prometheus with alert manager really for day two operations this is ready for you to to to get into production quickly so that you can alert to things as they occur and and really get a have your finger on the pulse of things with Ravana so I know it's not exactly what you're asked of you asked about the tools and those are great tools and they're built in day one oh great good stuff last question got one more for you guys you guys doing great so let's say that you know I'm interested in learning these tools you know I want to start with them right away you know what's your experience with you know how quick people can pick these tools up and start deploying them out you know in other words you know what's the learning curve you know we'll start with movies over there on PCF you know if a new person's new to the system what does it look like to get started PCF you know being fairly mature at this point all the tools have been straightened out and they work pretty well so getting PCF up and running on your platform is quite easy and there's extensive documentation on how to go about that the issue would PCF comes in because they're using a lot of proprietary tools you know if something does go wrong it's really difficult to debug it's difficult to troubleshoot yeah you need to have knowledge of Bosch which is the orchestration tool use for PCF and yeah and what is pretty much you know not used anywhere else outside of cloud foundry so those are some of the you know bumps that you face when setting up PCF in your environment okay great thanks for the answer huh all right have around the open ship side yeah so if you're putting up a bunch of cluster you have to put some effort I mean I'm talking about container platform your in-house custom so you have to invest anybody missing same here put some effort to bring up your cluster in configuring masters and work and all that stuff and if you talk about the user I'm using for example universe after functioning the cluster you can easily engage with open T because has a rich web interface why the base so most of the 80% of the past can be done in its rich web console see actually it has a really nice wave console you can navigate your application you can scale up scale down very quickly and you can change image tags for your deployments all can be done by our pet console I think it's a with zone 4 I missed something about the monitoring stuff I just want to add because in over ship the looking when you talk about logging the cost data I mean application logs you you have to put up a stack it's not by default we configure but manage management layer all the managers have to manage the API stack in order to serve and if it's not paid configured if you have to manage your logging and you have to upgrade so those stuff I need to be handle for great good answer Thanks and finally anthos what does that look like for setting up well first off because Fergie she referred back to a previous question stack driver is on the other side it's a it's integration from from GK on Prem integrates day one with stack driver and it pumps all of your logs and your metrics on out to your your stack driver environment which is pretty slick if you haven't used that at GCP so so so your question again was about training right or anything yeah what's the learning curve how long does it take for new people learn how to use it okay yeah honestly this is standard kubernetes it's it's a curated environment so it's set up in a more secure manner yeah the way that Google thinks it ought to be set up in the first place but it is standard kubernetes it qualifies as a kubernetes deployment by the CNF and so the the barrier to entry the the the training ramped up is really pretty short if you know kubernetes then you you'll feel right at home here fact yeah they opened the open ship folks would probably jump right in very quickly because it is it is kubernetes and it's and it's been it's also standardized so you're gonna be able to use the tools if you already used to so so Bosch dark Bosch is mentioned by moiez there and I really like Bosch I rent PCF for quite a few years still involved and so it's a great tool but it's it's difficult to to run on a day-to-day basis this this really is standard nodes and kubernetes they just happen to run in your vCenter environment great well thanks I'm gonna let you guys go and I appreciate everybody for hanging in with us and you guys have a lot more about ecolution and you see if we can come in and offer you a consultation for your company to find out what's the right product for you reach out to us and we'd be happy to come in and do that

Channel: Techolution

Views: 1,904

Rating: 4.652174 out of 5

Keywords: anthos, google anthos, hybrid cloud, gcp, google cloud, PCF, Openshift, cloud transformation, google anthos pricing, anthos pricing

Id: 7B0jOuZq_DY

Length: 18min 56sec (1136 seconds)

Published: Fri Oct 11 2019