Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [mosquitto-dev] Advice needed (long, sorry)

If you're really concerned about price, look into reserved instances. EC2 can be bought spot, demand, or reserved, and if you're willing to commit and put down some money upfront, you can save 40-50%. Just read about it on the AWS site, a surprising number of people don't know about this. You can go even cheaper if you buy an RI off of someone else rather than directly from Amazon, which they also allow via their marketplace.

On Thu, Feb 12, 2015 at 1:12 AM, Manuel Domínguez Dorado <manolodd@xxxxxxxxx> wrote:
Hi David, I appreciate your reccomendations.

I agree with you that we are not going to save time. But as we already have our servers (not a datacenter, but a home-based datacenter) we thought that beybe we could take advantage of them using them only for a few months (a year max) decreasing costs as we already have this proliants. At the same time we could use this period to analyse growth patterns, users preferences... maybe we are wrong, of course. This is something that we have not completely clear.

Think that we also has to spend money during this phase to put the project in the market: marketing, launch campaign... and to create our company (it is expensive in Spain). But we know that this should not be a long term solution (sure you are right, we know this).

Also, we really do not think that a simple t1.micro is enough. Our current deployment is composed by 8 VM; none of them the kind of instance that a t1.micro is, but higher (we did some proofs before configuring them as they are configured). So we think we need t2.medium and m3.large instances (some of both) only to to start. At current AWS pricing this is about 600-700$/month. So... it is nor a great neither a small invest. We are doubting at this point. We will do a deeper analysis on this.

Thanks.

PS: 50 msg/hour is not the total number, but 50msg/hour per client, so it is the aggregated traffic what matters. And we do not know how it is going to grow and the number of concurrent connections we will have. Somethim more to think :-)
PS2: I'll contact you, of course.


2015-02-11 23:46 GMT+01:00 David Albrecht <david@xxxxxxxxxxx>:
My .02 on this -

I worked on an operations team for several years at a high-volume app performance monitoring company (you can look me up online if you want to know which one - I don't want to put the name here to stay off of Google). We monitor almost a billion app-instances across xx million mobile devices globally. A few suggestions on the ops side:

You aren't going to save any money, and you certainly aren't going to save time, racking your own servers. Just go to AWS, doing a hot-cutover is extremely difficult to do without downtime even for pros, and it's not value-added work - try to avoid it. Even better: find a PaaS provider that will run all your infrastructure components (e.g. mosquitto) and pay them to do it. PaaS might be more expensive upfront, but the sheer hassle of wearing the pager, dealing with power supply failures, designing a network topology, physically racking the machines, and driving to datacenters isn't something you need to do.

50 messages/hr isn't going to break even a small server, hell you could probably run that on an EC2 t1.micro (an instance type so small they don't even make them anymore) without a problem. 

Email me off-list if you want more advice, I'd love to hear more about what you're doing.
DA

On Wed, Feb 11, 2015 at 2:07 PM, Oegma2 <oegma2@xxxxxxxxx> wrote:
Hi Manuel, I agree with Stefano - done a lot of detailed work 

Just though of an option - never tried it but maybe worth while looking into using Redis with more than one mosquitto instance - using HAPraox as load balancing you can run say 3-5 mosquttos with a redis layer behind mosquitto ....so in short HA --> Mosquitto --> Redis <-- other sub/pub on server side.....



On Wed, Feb 11, 2015 at 11:47 PM, Stefano Costa <stefano.costa@xxxxxxxxxxx> wrote:

> Il giorno 11/feb/2015, alle ore 22:15, Manuel Domínguez Dorado <manolodd@xxxxxxxxx> ha scritto:
>
>
> Could you give us some advice on how to improve these numbers at least to be able to reach our 100K users? We don't know, but maybe 20K concurrent connections are enough to support 100K users that are not connected at the same time.

Hi Manuel, it seems to me that you studied all details very well, good architecture as a starting point. But in order for people here to give useful advices I suspect that an important detail is missing: apart from "connected" and "existing" users, what's the messages rate you expect to have per user / per second (or minute, hour) under normal conditions and maximum?
_______________________________________________
mosquitto-dev mailing list
mosquitto-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/mosquitto-dev


_______________________________________________
mosquitto-dev mailing list
mosquitto-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/mosquitto-dev



--
david r. albrecht
prefiat llc | cloud powered hardware

_______________________________________________
mosquitto-dev mailing list
mosquitto-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/mosquitto-dev



--
---
Manuel Domínguez Dorado
ingeniero ARROBA ManoloDominguez PUNTO com
http://www.ManoloDominguez.com

_______________________________________________
mosquitto-dev mailing list
mosquitto-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/mosquitto-dev



--
david r. albrecht
prefiat llc | cloud powered hardware

Back to the top