Friday, 30 October 2009
Hostage to fortune
First things first. I'll be using the NIST definitions for cloud computing - check them out, they're good and they're vendor-independent. [One beef I did have with the speakers at the Cloud WF was that they all insisted on giving us their own definition of cloud computing. We really should be over that by now... Particularly when they all mentioned the Internet and then a number went on to talk about private clouds.]
Let's have some initial assumptions:
i) IaaS will become more interoperable and portable - either provider-supported through the use of standard APIs (check out http://www.occi-wg.org) or by default through meta-cloud providers reverse engineering closed APIs.
ii) PaaS and SaaS vendors will have a big question to answer around the granularity of the services that they offer.
iii) Consumers will have some serious thinking to do with respect to the amount of lock-in (and subsequent pricing consequences) they are willing to endure.
So in my future IaaS will become seriously commoditised with consumers able to switch loads or other basic IT needs as and when necessary through the use of meta-clouds or other mechanisms for managing multiple cloud providers. I think that's a given. [I'm not going to talk about private or community clouds much in this post, let's just assume that most internal IT systems will be delivered by either private or community cloudy resources - let's face it, there's not much that won't be virtualised in 5 years time other than the obvious usual suspects, y'know those guys still running Cobol on legacy kit...]
The PaaS and SaaS space is much more interesting. In an ideal world, these kinds of providers would completely open up and offer very granular services, presumably charged per transaction or subscription, that consumers could use on a per-service basis from outside of the provider environment. Enabling SOA via cloud services. That would be good. What I fear is that PaaS providers in particular will be very close minded in their thinking and actually encourage the PaaS lock-in that has many cloud commentators (including this one) worried. Why would they do this? Well, once a consumer is effectively locked-in there'll be every temptation to start upping the prices - as long as the pain to the consumer is less than migrating away from the PaaS it's a definite win for the provider. Ah, but competition will prevent this I hear you say. Well, only if the competition isn't doing the same thing!
So that's my view of how the future will pan out. Anyone care to share theirs?
Friday, 23 October 2009
Cloud World Forum
Highlights and interesting tidbits from the event:
o Kate Craig-Wood of Memset, Intellect and the BCS is now co-leading the technical architecture stream of the Cabinet Office data centre consolitation work
o Asite are a public cloud service that have apparently obtained HMG accreditation for use by the Environment Agency. Unfortunately the presenter left before I had a chance to quiz him on the accreditation aspect!
o Lots of good presentations from the likes of Gartner and BT and some interesting panel sessions, particularly interested in the Gartner research that showed security was still the leading concern with organisations yet to adopt cloud computing. Also interesting that the main drivers for those organisations that have adopted cloud computing were cost and functionality. Who'd have thought it? ;-)
o If you have an interest in collaboration then certainly check out www.huddle.net - collaboration tools, video conferencing etc all in one user-friendly cloud-based offering.
o BT's virtual data centre is an interesting proposition - they do not run VMs for more than one customer on a physical blade. Of course, from a paranoid perspective you may still have de-commissioning concerns when the blade is returned to the wider resource pool. Not dug into the real low-level details here.
o Mimecast have released a Forrester Consulting report into the "total economic impact" of their solution. Yes, the report is specific to Mimecast, however the methodology of the report is of interest and it's useful to have a (vaguely) independent, albeit funded, report showing a detailed ROI argument for a cloud-based service. The report should be downloadable from the Mimecast web-site but I don't think it's there yet.
Downsides:
o Terribly dull presentation from VMWare, Cisco and EMC. Everybody else talking about business benefits, these guys droning on for a long time about IT and infrastructure issues. Bored everyone to tears. Content was actually not bad from a technical perspective but was wrong for the event and the delivery was way too dry. [Example of the problem with the presentation, when talking of moving to cloud services "...got to start with server virtualisation" - well, only if you're talking IaaS and I'd personally start with identifying what you want to do from a business perspective!]
o Still a general ignorance with respect to security - lots of mentions of it during the day but no real understanding of how to manage risk in a cloud environment. [One panellist even described escapes from VMs as 'a bit of a myth' - a bit problematic given that exploits have been published which do just that...]
o Slightly disappointing presentation on cloud security from Cryptocard which was basically yet another demonstration of using Cain and Abel to intercept passwords (*yawn*) and an overly broad statement that 2 factor authentication solves all authentication issues in a cloud environment. Yes, they would say that being as they sell 2FA solutions but it's blatantly not true!
Overall - good event, will definitely try to attend next year's. The attendees were left with the feeling that cloud computing is here, is real and is delivering benefits to the early adopters.
Thursday, 15 October 2009
Resources for the busy security pro...
Firstly: http://archives.neohapsis.com/
There are loads of security mailing lists - the site above is a convenient method for keeping track of the most useful ones. I'd recommend their Yesterday, Today, Full-Disclosure and DailyDave archives. There are other aggregators but I've been using this one for years and I'm a loyal soul...
Secondly: http://www.monkey.com/~jose/secblogs.html
As with mailing lists, there are loads of security blogs and loads of blog aggregators. I tend to use the one above as it aggregates blogs I'm interested in and provides a manageable number of links per day - I don't feel overwhelmed by the sheer volume of posts!
Hope you find them useful. If you have any other resources that you think would help a busy security guy keep up to date (in a quick and manageable way!) please add some comments.
Thursday, 1 October 2009
Cloud Security Summit
If you're interested in cloud security, my web-cast can be found here:
http://www.brighttalk.com/webcasts/5688/play
If you have any questions or want to leave any feedback, feel free to comment :-)
Monday, 28 September 2009
CloudCamp London 5
After the talks we broke into Vendor Tracks and Open Space discussions - one of the attendees suggested a session around security which I volunteered to moderate. I've written up these discussions and the write-up is shown below. If any of the attendees feel I missed anything out or have misrepresented the conversation please feel to comment or drop me an email. Enjoy!
London CloudCamp #5 Open Space – Security (Room 3)
Chatham House rule applies!
i) Public sector in the Cloud
Discussion began with whether the Public Sector would adopt cloud due to their security requirements. It was noted that the UK Government is planning a G-Cloud as part of the Data Centre Consolidation Strategy – this was also a recommendation into the Carter Review (Digital Britain). Attendees were also pointed towards the blog of John Suffolk, the HMG CIO – http://johnsuffolk.typepad.com. It was thought unlikely that public clouds would be suitable for processing of protectively marked information (i.e. RESTRICTED and above) – although it may be possible to use them for storage and transport if data is encrypted and decrypted on-premise. There was thought to be more likelihood of public clouds being used within local government where security requirements are less stringent due to their data typically being at PROTECT. The main sticking point from a security perspective was currently thought to be around the lack of assured products to support domain separation.
ii) Certificate based authentication
There was a discussion as to whether cloud computing made it difficult to use server certificate based authentication due to the need to tie certificates to domain names or IP addresses. It was not thought to be a problem with IaaS (where this can be controlled by the consumer – if the right technologies are used). Thought to be problematic with PaaS and SaaS.
iii) PCI-DSS and ISO27001
There was a question as to the overlap between PCI-DSS and ISO27001. The group believed that there is significant overlap between the two standards but that PCI-DSS was more prescriptive and so compliance with ISO27001 did not mean compliance with PCI-DSS. PCI-DSS has specific requirements around handling of cardholder data, vulnerability assessments etc that are more granular than those within ISO27001. The recent blog post including the AWS statement that it was not possible to be completely PCI-DSS level 1 compliant using only their EC2 and S3 services was discussed. It was noted that you can simply hand off payment processing to a third party payment processor or keep such processing in-house. It was also noted that there is a separate PCI standard covering the development of payment processing applications.
iv) Privacy
We had a brief discussion around privacy legislation – one of the attendees noting that Germany is about to enact a notification law such that any organisation suffering a data breach must notify all affected customers (either by individual letter or by taking out a 2 page advert in a national newspaper).
v) Use of cloud resources for illegal purposes
We had a particularly interesting conversation around the use of cloud computing resources for illegal purposes – for example the distribution of cracked software keys. This discussion was illustrated through real examples of previously identified instances of such activity. This does raise interesting questions about whether cloud providers should be monitoring for such activity or whether they, like telco's, should act simply as carriers.
vi) Data leakage
The idea that data could be split throughout the cloud to make re-constitution more difficult was discussed. It was thought that this was already one of the benefits of cloud computing – should a service provider lose a disk, it is most likely to contain fragments from a number of clients rather than a substantial chunk of a single organisation's data.
Miranda Mowbray's obfuscation tool and the Vanish tool (Washington State University) were mentioned as being of interest to those looking to keep sensitive data under control. Both noted as being primarily of academic interest at this time.
vii) Virtual Desktop Infrastructures
There was some discussion of VDI in the cloud. Noted that the public sector may “browse-down” from a more sensitive domain to a lesser domain, e.g. to offer Internet access via terminal services but that "browse-up” was frowned upon.
viii) Security Benefits
It was thought that the cloud model can offer some security benefits – e.g. Increased/improved security monitoring, patching, security expertise and physical security. Likely to be of more benefit to SMEs but could also be of benefit to larger organisations (most of whom should already have invested in the necessary functions).
ix) Security as a Service
The prospects of security as a service were discussed. It was noted that businesses such as MessageLabs have been doing this for years! Security filtering in the cloud is a valid service. Could also expect to see identity providers in the cloud in the future.
Monday, 21 September 2009
Cloud miscellany
But back in the real world it's been interesting at work with respect to how many of our current bids and engagements are now considering delivery, at least in part, via cloud computing models. There seems to be a real shift to treating cloud computing as just another part of the delivery model a la outsourcing, right shoring etc. I have to say that I thoroughly approve of this change - technology for the sake of technology, or even change for the sake of change, is never a wise thing unless of course you're in a particularly bad place and are due a change in luck! What we do have to remember is the potentially game-changing nature of certain cloud computing characteristics - in particular increased agility - which means that we need to be careful not to limit our imaginations to doing just the same things but in a different way. Don't forget to think different, but most importantly don't forget to think!
Friday, 21 August 2009
BrightTALK Cloud Security Summit
There are some well-known and well-respected figures presenting during the summit - details of the other presentations and presenters can be found at:
http://www.brighttalk.com/summit/cloudsecurity
Come along, I'm sure it'll be fun. I may even have thought of some interesting voting topics by that point as well - I'd welcome suggestions if anyone out there would care to volunteer some?
