Monday, 30 April 2018

AWS revenue up 49% YoY - Almost $20B annual revenue

http://phx.corporate-ir.net/phoenix.zhtml?c=97664&p=irol-presentations


AWS are making money hand over fist. Very, very clever....buying commodity hardware at massive scale then wrapping open source around it and selling services at significant profit. AWS account for 73 percent of Amazon's $1.93 billion in operating income

Scott Galloway's predicts in his book "The Four" that Amazon will be the first trillion dollar company by market cap.

Thursday, 12 April 2018

SDWAN: Velocloud, Cisco invested in them then VMware bought them

Velocloud, Cisco invested in them then VMware bought them....interesting that Telstra partnered with them:

All it really is a nice GUI on top of open source Linux with open source Quagga for the routing functionality. Quagga is basically a clone of IOS such that you can run IOS's sh ip bgp vpnv4 vrf <name> neighbor <peer_ip> received-routes and produces exactly the output as from IOS. Control plane is completely in the cloud. 

Checkout the following deck that reviews SDWAN economics:

Cisco bought Viptela for $610m in May 2017, VMware bought Velocloud for undisclosed sum in November 2017. So it looks like NSX will be used for local networking within vSphere and Velocloud will be integrated with NSX for SDWAN branch networking. Hence VeloCloud is a safe bet (as long as the price is right!).
Notes on internet based SDWAN QOS:
Any QoS will only apply to your own traffic. The same would be true if you rolled your own VPN. The big difference is that once your traffic hits the internet, it can be dropped. This is not the case with an MPLS circuit with proper QoS. Your traffic isn't competing with anything.

Incoming traffic operates the same way. With MPLS circuits, its private. Its just your traffic. With services using the open internet, you can be DDoS'ed to oblivion, and you will never see 1/50th of the SD-WAN traffic you are supposed to be getting. The ISP will simply drop all of it.

That being said, one can use open internet VPNs to connect locations, and its good enough. More importantly, its a lot cheaper. The only caveat is that sometimes you run into situations where your services are impacted due to the nature of the connection. MPLS circuits avoid this... but they are much more expensive.

SD-WAN services just a better form of management. If you don't have a team configuring your routers, keeping the configuration secure, ensuring everything is done right... then this is an easy way to get it done. Of course, your security and reliability also depends on the SD-WAN provider's security and reliability.

Thursday, 5 April 2018

VMware Self Service Automation - VMware vRealize Automation vs open source vOneCloud

https://www.vmware.com/products/vrealize-suite.html
https://www.vmware.com/files/pdf/vcloud-suite-pricing-packaging-whitepaper.pdf;

vRealize Automation Architecture:

Self Service Catalog:


VMware vRealize Automation 7 Details
FeatureInformation
ProductSoftware for managing and provisioning cloud infrastructures.
ManufacturerVMWare (www.vmware.com)
PriceThe advanced license of vRealize Automation for managing 25 VMs is available from around $6,245. The enterprise license starts at $7,745.
System RequirementsvRealize Automation has very exacting hardware requirements in part; the individual components are subject to various requirements, which can be found in the product data sheet .
Technical Datahttps://pubs.vmware.com/vra-70/topic/com.vmware.ICbase/PDF/vrealize-automation-70-reference-architecture.pdf

http://vonecloud.today/ 
Easy to use "cloud like" user interface that allows self service and CLIs/APIs:
https://docs.vonecloud.today/2.2/simple_cloud_deployment/interfaces.html

Tuesday, 3 April 2018

Apple Plans to Use Its Own Chips in Macs From 2020, Replacing Intel

https://www.bloomberg.com/news/articles/2018-04-02/apple-is-said-to-plan-move-from-intel-to-own-mac-chips-from-2020
Predictably Apple is planning to move Mac computers from Intel CPUs to their own ARM based CPUs manufactured by TSMC.

Their CPUs are good enough in terms of performance and would enable Apple to more tightly integrate hardware and software, have better battery life and significantly lower cost.

Apple has changed CPUs on Macs a number of times:
Such a change can be painful as applications will need to be rewritten to use the new instruction set and perform well. In the past they have used dynamic translation which converts from one instruction set to another, yet this has significant performance implications.

Lets compare performance of Apple's ARM CPUs and Intel CPUs:

iPhone X with A11 CPU performance:

Macbook Pro 15" with Intel  i7-7920HQ CPU performance:

For typical end users performance with Apple ARM based CPUs will be perfectly fine, the single core performance is already on par with Intel CPUs and in most cases the multi-core performance is good enough for 95% of users. Yet performance will not be competitive for high end Pro users such as developers running multiple VMs, 4K video editors, AI processing and other processing tasks that demand high end multi-processor performance. Apple have a trend where they are no longer serious about Pro Mac users.

Yet I believe this change is generally good move for Apple as it will have a unified architecture. This will have the potential to use an iPhone as your real computer:
See more details below: