Disclaimer: I recently attended the EMC Forum 2013 – Sydney. My flights and accommodation were paid for by myself, however EMC provided meals and some swag. There is no requirement for me to blog about any of the content presented and I am not compensated in any way for my time at the event. Some materials presented were discussed under NDA and don’t form part of my blog posts, but could influence future discussions.
In this post I’d like to touch briefly on a few of the sessions I went to and point you in the direction of some further reading. I’m working on some more detailed content for the near future.
Backup and Recovery – a Peak under the Hood of a Service Provider: Telstra
Kevin Kennedy is Director of Cloud Engineering at Telstra, and spent some time setting the scene as to why cloud is a good option and how cloud could be used as both a DR and backup / recovery solution. He suggested that we, in the IT industry, should be asking ourselves whether we’re more about the technology or the information. It’s a problem that exists in a number of IT departments, and one that needs to be revisited regularly to ensure that IT is really adding value to the business. As this session was a little more business-focused than technical, a fair bit of time was spent on ideas like agility and why cloud is good. Nonetheless, brief mention was made of VMware SRM and EMC RecoverPoint. It was nice to see that, at least on the surface, Telstra aren’t doing anything too earth-shatteringly different to most Iaas / BaaS providers.
How Cisco IT Delivers Next Generation Cloud Services
Mike Paranihi is Director of IT and Service Owner of DC Facilities and DCaaS at Cisco. Mike’s was one of the more entertaining presentations of the day (possibly because I’m nuts deep in cloud transformation projects at the moment), and I suggest if you have a chance to listen to him or meet with him that you take the opportunity. Mike is responsible for 34 facilities across the globe. By the numbers, Cisco IT supports over 300 locations in over 165 countries with over 65000 employees. 80% of their compute is virtualised and they’re aiming to get that to 95%. Use VMAX and VNX for T1 and T3 storage, Atmos for cloud storage, Vblocks and, overall, they use 96 EMC arrays with 21PB of raw storage (13PB allocatable).
Challenges? They’re always getting asked to cut the budget, do more with less and continue to maintain or improve service levels. The also need to be able to service internal clients who have evolving and rapidly changing resource requirements. So they created CITEIS (Cisco IT Elastic Infrastructure Service) about 4 years ago. Mike then went on to clarify the taxonomy he was using (DCaaS – facilities; IaaS – compute / storage / network; PaaS – the platform sitting on top; and SaaS – software, eg salesforce.com). CITEIS is predominantly an IaaS and PaaS play for Cisco IT.
You need unified fabric, which enables unified compute (use VMware internally) – this is Generation 1. They implemented process automation and self-service opportunities. Starting to investigate how they can cloud-burst, and begin to leverage capacity from other cloud providers. The portal is key to the success of this with the end-user (without the end-user there is no point – something we forget from time to time). Process automation (in CITEIS v5) now works at the hypervisor layer with both VMware and openstack.
The right security models are key to good multi-tenant deployments. Cisco heavily leverages the Nexus 7k VRFs, trusted zones and large subnets for this. Provide granular L2 security with VSGs. Primarily using EMC NAS for CITEIS storage.
Interestingly, they offer internal clients an “Express” option, that comprises 2 VMs for 90 days for use in PoCs and so forth. They can then choose to move those VMs into a larger vDC (virtual data centre) if required (minimum of 3 months consumption, guaranteed resources). vDCs are elastic and built via catalogue. Managed and self-managed options are also available. The big benefit of this approach for Cisco has been a reduction in provisioning time from 6 – 8 weeks down to about 3 – 5 days. They would like to get this down to 15 minutes but, ironically, are still working through issues with network management controls and some security procedures.
If you want to do PaaS well, you really need to understand exactly what services you want to offer via the service catalogue. You also need to understand which ones are high-frequency, and what are the foundational elements of these offerings in order to properly provision the underlying infrastructure. Cisco created IT organisations that revolved around service delivery, not function delivery. They’ve been benchmarking the TCO of their PaaS offering internally and are seeing real cost savings. Working towards having everything as a service available.
Also investing heavily in programmable infrastructure. Also now looking at how to leverage private / public / hybrid cloud while considering multi-tenancy and security requirements. The next step is application driven, where applications give instructions, based on SLAs and business rules, to the PaaS and IaaS layers to provision suitable bandwidth / compute / IOPS (eg an app getting smashed at end of quarter).
In Part 3 I’ll be looking at the ViPR and VNX sessions I attended.