1 Big Data. Better progoses ad quicker decisios. With Big Data, you make decisios o the basis of mass data. Everyoe is curretly talkig about the aalysis of large amouts of data, the Big Data. It gives compaies the opportuity to make quicker ad better decisios. However, there are various challeges ivolved with the gatherig, savig, ad evaluatio process of the data. Therefore, secure data filig is eeded, as is IT security par excellece. The Quick Guide Big Data addresses topics that are relevat ad futureorieted for busiesses. Six experts report o hadlig mass data ad the opportuities associated with the aalysis of Big Data. YOUR CONTACT PERSON. Ralf Korad Offerig Maager BI & Big Data Facebook: Iteret: CONTENTS: Trasformig mass data ito successes How the data explosio ca be successfully used. Big data: drawig the right coclusios Frak Niema, Pricipal aalyst at Pierre Audoi Cosultat (PAC) Why classic tools ad techologies are t eough. Big busiess thaks to Big Data Heiko Hekes, Aalyst Maager at TechCosult How to make kowledge more tagible. Big data istead of big cofusio Holm Ladrock, Seior Advisor at Experto Group How Big Data icreases the level of quality. No big data without cloud computig Markus Vehlow, Parter at PriceWaterhouseCoopers (PwC) What role cloud computig plays. At the press of butto: better predictios ad faster decisios Hage Rickma, Director Sales at T-Systems What the future of Big Data looks like.
2 Trasformig mass data ito successes The gigatic 12 terabytes which are geerated through Twitter feeds every day, five millio global share trades per secod, ad more ad more videos, photos ad other ustructured data. These facts speak for themselves clearly eough the Big Data tred is takig shape. However, it s ot just a matter of gettig a grip of this mass of data, but also about its speed ad variety. If compaies cotrol these three factors, iteral aalytical expertise is crucially importat i order to fid the correct key figures, that is, to recogize patters, sigificat poits ad cotexts. Here is the good ews: There are already compaies which have successfully deciphered the Big Data code ad which have got a grip of the data explosio such as Macy s. With reveue of almost USD 30 billio ad a etwork of 800 locatios, Macy s is cosidered the largest store operator i the USA. Nevertheless, the 180,000 employee-strog compay maages to ru a daily price check aalysis of its 10,000 articles i less tha two hours. That is all the more remarkable sice Macy s also has high price volatility withi the compay. This meas: Wheever a eighborig competitor aywhere betwee New York ad Los Ageles goes for aggressive price reductios, Macy s follows its example. If there is o market competitor, the prices remai uchaged. Thus there are aroud 270 millio differet prices across the etire rage of goods ad locatios. The fact that this price aalysis is possible at record speed uthikable before Big Data aalysis is somethig that Macy s CIO Larry Lewark owes to switchig its existig ifrastructure to a cloudprovided software solutio from SAS ad to the use of i-memory techology. I this way, Macy s ca eve adjust its prices several times o the same day to react better to local competitio. calculatios. However, i order to use these to examie the effects of the market o the total risks of the bak, IT recetly required up to 18 hours. Thus a prompt reactio to ewly arisig market risks was impossible. Thaks to Big Data aalysis, the experts at UOB have ow succeeded i cuttig the risk calculatio time dow to oly a few miutes, ad i takig accout of rapidly chagig parameters i the complex aalysis almost i real time. Whereas up to ow risk aalysis has sometimes bee perceived as a irritatig task that had to be performed at the behest of the supervisory authorities, UOB ca today use the istrumet to assist its operatig busiess. This allows UOB to test tradig strategies i advace ad to assess the likely effect of ew market evets more quickly. Bak cotrols the risk The Big Data success cocept for the Uited Overseas Bak (UOB) i Sigapore is also a high performace aalysis solutio from SAS, combied with i-memory techology. Here the skillful aalysis of large data volumes has made a sigificat cotributio to the 45 percet rise i the share price of UOB i recet years because the large fiacial istitute i south-east Asia is a master of risk maagemet. I detail: At UOB, the risks are divided betwee 45,000 differet fiacial istrumets ad are set by usig more tha 100,000 market parameters such as prices, deadlies ad due dates. The calculatio of the overall risk assumes aroud 8.8 billio idividual, highly complex value at risk Iteret resource: 2
3 Big data: drawig the right coclusios PAC Aalyst Frak Niema talks about the specific opportuities that the big data pheomeo offers busiess. Mr. Niema, the Iteret ad cloud computig have made big data aalytics a reality. But what are the prime drivers of this tred? Big data is t just hype. We are t tryig to create a market out of othig. Compaies have to recogize the existece of a objective eed, drive by the icreasig digitizatio of processes. Moreover, applicatios are geeratig data that we should ot simply store, but also aalyze. Ad, much more data is available from sources that eterprises had ot cosidered util ow social media cotet, for example. Drawig the right coclusios from that data is key. It eables you to accelerate busiess decisios ad implemet the ecessary chages i the way we maage customers, products ad processes such as customer service. Why are t covetioal busiess itelligece tools ad techologies up to scratch? It s ot simply a case of lettig existig aalysis systems loose o ew data. First, that data must be itegrated, made available ad made ready for aalysis. With social media cotet, for istace, that is far from easy. Data from this kid of source is ot structured i the same way as data stored i a database. Data maagemet ad data itegratio are particularly challegig tasks. This meas that the huge volume of data, with its diverse formats ad chaels, requires ew systems, fresh thikig ad reorgaizatio. Frak Niema Pricipal Aalyst at Pierre Audoi Cosultats (PAC) What ca IT service providers do to facilitate uderstadig? They ca t close the gap completely, but they ca facilitate dialogue betwee IT ad specialist departmets. That aloe would be a big step i the right directio for compaies. I reality, service providers do t go ito compaies with a big box of solutios ad istall a ew server, for example. Istead, they offer services such as workshops, to help develop the ecessary uderstadig of the opportuities ad challeges of big data. Reorgaizatio suggests that it is ot purely a techical exercise; people also have a importat role to play. To what extet is it liked to how well the IT departmet ca uderstad ad use the laguage of specialist departmets? That s exactly the poit: fidig a way of combiig techical expertise with busiess process skills. It s a old debate, but with big data aalytics it has grow cosiderably i importace. Specialist departmets already have high expectatios of data aalysis, ad big data will raise expectatios eve further. Specialist departmets ca fid it difficult to express their requiremets i terms of fuctios ad services i a way that is comprehesible to IT professioals. Ad it s the same i reverse. Iteret resource: 3
4 Big busiess thaks to Big Data Heiko Hekes, a aalyst maager with TechCosult, kows what exactly makes up the added-value of Big Data for which sectors. Mr. Hekes, the process of aalyzig large data volumes has already bee established at cosumer goods maufacturers ad i the fiacial sector. From your poit of view, are there other sectors that ow eed to follow with Big Data aalyses? I priciple, every sector with large data volumes is required to do so, somethig that will soo affect virtually all compaies. Other sectors with steadily icreasig requiremets with respect to the itelliget aalysis of huge amouts of data are isurace compaies ad, i future, more tradig compaies, too. The expoetially icreasig data growth i today s society is closely coected to corporate big data. Big Data meas makig kowledge more readily available ad helps to greatly improve the degree of plaig i compaies ad thus create measurable competitive advatages i the short-term. Ca you give us a example? Sure, a example of this would be the utilities idustry; the keyword here beig smart meterig. Utility compaies ca obtai a cosiderably more detailed view of requested peak loads ad types of users. As a result, they ca offer their packages i a way that is more specific to target groups as curretly the sector has to lump all users together. Savigs ca be made o process costs through smart meters that commuicate automatically because maual readouts o loger have to be carried out. The user also obtais more trasparecy ad eergy efficiecy due to data always beig up-to-date. A icreasig umber of specific requests are reachig the CIO from his specialist departmets. Why is this? Because fudametal maagemet requiremets ca be solved this way: Marketig obtais i-depth kowledge about the relevat target group s wishes ad requiremets. Procuremet ca improve the respose speed to raw material requiremets ad fluctuatios thaks to real-time aalyses. I tradig compaies, this area has a particularly high ifluece o the margi. Models already established o the basis of a i-memory database e.g., SAP HANA TM already assist procuremet professioals i optimizig the procuremet processes here. I marketig however, it is possible to optimize the market approach through up-to-date use typologies that correspod to the busiess models. That s the theory i practice there are still some obstacles to be overcome so that compaies are actually able to utilize this added value. The persoal compoet plays, as before, a crucial role i the success of the busiess particularly i Procuremet. Heiko Hekes Aalyst Maager at TechCosult ito aythig useful. Fault also lies with the eormous rise i mobile devices o which live work is icreasigly carried out. However, these are rarely, if at all ivolved i the process ad iformatio chai. Here, there is ofte a lack of experiece values or best practices. Solutios are ow available o the market. I view of fast-paced developmets ad users becomig idepedet i combiatio with techology, this is difficult for compaies to set up by themselves the leadig had is missig i topics such as Security & Compliace vs. CoIT ad/or ByoX. Who will be the first to overcome these obstacles? Compaies with a iterest i IT will most likely be able to take advatage of these opportuities. They have the ability to trasfer busiess processes ito IT-related processes. I doig so, they ca also embrace competitive advatages more quickly. Geerally, this tred ca o loger be stopped: Big Data will icreasigly evolve ito big busiess through compaies beig able to esure cosiderable competitive advatages by meas of thought-out ad fast data aalyses. However, flexible ad scalable meas of provisio for this must ot be forgotte: oly cloud computig, usually i coectio with service-based architecture, establishes the basis for the collaborative corporate processes that are essetial for this. Why is that? New Big Data tools are curretly grapplig with a extreme icrease i data which is rarely budled ad is frequetly produced ad stored without defied processes. I the last few years i particular, there was a huge amout of data collected via BI tools that obody coverted Iteret resource: 4
5 Big data istead of big cofusio Why weather data is likig football games ad supermarkets, ad how tweets ca impact the supply chais of cadymakers: Holm Ladrock, Seior Advisor at Experto, talks about big data scearios ad log-term approaches to aalytics. Mr. Ladrock, Experto recetly coducted a study o how eterprises curretly develop effective big data strategies. What was your most importat fidig? To aswer that questio, we must keep i mid that big data withi the cotext of a major corporate eterprise ivolves a much more complex sceario. I other words, more is required tha just makig ivestmets i ew appliaces or storage systems that are iteded to speed up existig data warehouse operatios. That kid of approach is ot a strategy. What eterprises should do ivolves applyig big data techology to develop scearios that have ever bee see before. That meas likig up busiess admiistratio with geographical data, for example. Today completely ew ways of thikig are required. This is where the idea of big data is rooted everythig else is just a reflectio of covetioal busiess itelligece scearios. Where are geuie big data aalytics beig used today? I the major supermarket retail chais. Goods are delivered to the supermarkets based o data that icludes weather forecasts ad the game schedule of the atioal football league: If a hot ad suy weeked is expected, the stores eed to have plety of beer ad hot dogs o had for barbecues. That is busiess itelligece for ivetory maagemet. But what happes if supermarket sales uexpectedly slump, ad there is hardly ay turover durig a weeked of high-profile football games? That is a very iterestig questio that could be aswered by aalyzig all of the data available which would certaily result i a much more accurate forecast. I what way? I am talkig about predictive aalytics. For example, the employees of a food compay ca search through social etworks to Holm Ladrock Seior Advisor at Experto Group fid egative commets posted about their products. This helps the compay idetify possible issues ad eables it to take corrective actio quickly. Big data aalytics could automate this process by aalyzig millios of data sets from a large umber of social etworks. This would give the compay cosiderable competitive advatages both ow ad i the future. Whe aalyzig iformatio from social media, data privacy must be guarateed by makig sure that all data is aoymous. For example, it is ot importat to kow who said this or that about a particular product, but discoverig ay regioal treds or issues ca sigificatly impact busiess. How ca eterprises achieve a high level of quality i their big data aalyses? New processes like Hadoop are importat. They are helpful whe preparig structured, semi-structured ad ustructured data for aalysis. Ad to map the complexity required for big data, ew algorithms are also required as well as people who will ask ew questios. How ca providers support busiess eterprises i acquirig these abilities? They ca do so by developig big data scearios. They must provide solutios offerig beefits that perfectly match the eeds of specialist uits ad their daily tasks. The example I just metioed for supermarket retailers would certaily ot satisfy specific requiremets i maufacturig. Furthermore, eterprises eed to trai data aalysts who will be able to trasform the theory ito real-world processes. Thus a holistic approach is ecessary ad ot just for the short term. Why? Is t immediate ad short-term actio the key to success i this very dyamic sector? Big data will certaily be a topic requirig itesive attetio over the ext 10 to 15 years. It is ot a short-term tred ad thus should be give some careful ad strategic thought. Big data is a disruptive techology that ca deliver cosiderable market advatages to those busiess eterprises that kow how to employ it effectively. Oe way of promotig big data scearios could be the use of dyamic cloud computig ifrastructures that provide big data aalytics as a service. For example, this could iclude services like scaig data from various sources accordig to sematic uits or keywords. Iteret resource: 5
6 No big data without cloud computig To effectively aalyze big data, it has to be collected, stored, cleaed, aalyzed ad illustrated graphically. Ad that is just the begiig. Aalyzig persoal iformatio from sources such as social etworks ca lead to a violatio of compliace regulatios ad laws. Markus Vehlow, Parter at PriceWaterhouseCoopers (PwC), o how compaies ca avoid these problems ad how cloud computig ca help. Mr. Vehlow, a lot of compaies are startig to work with big data i close coectio with cloud computig. What does that mea i terms of compliace ad data privacy? I geeral, the same legal framework applies. The existece of a ew IT tred is ot eough to chage the law. The same goes for big data ad cloud computig. It becomes a legal issue whe a flood of data is beig collected, whe persoal data is compiled ad ca the be traced back to actual people. At that poit, mechaisms eed to itervee that secure the use of big data for busiess purposes but also provide sufficiet data privacy. You mea techology? You ca use techology to make data aoymous or apply pseudoyms. But we see time ad agai that compaies first eed to employ other orgaizatioal measures i order to guaratee the effectiveess of these processes. What type of measures? Data classificatio. Highly regulated idustries i particular, like the fiace idustry or the healthcare ad pharmacology sectors are a good example of how to use data classificatio. Because of curret regulatios, may compaies take stock regularly ad coduct regular risk aalyses o data or eter these ito a risk register. This geerates results such as: Classes 1 to 2, iformatio that is ot so critical, ca be placed i a public cloud, classes 3 ad 4, such as persoal or busiess-critical data, defiitely eeds to be stored iterally, ecrypted or made aoymous. Ufortuately, ot every compay that uses big data ad the cloud is that far advaced. Markus Vehlow Parter at PriceWaterhouseCoopers (PwC) e.g., usig social media data, ca really help compaies make the right decisios. But these aalyses always eed to have a purpose ad a goal. Just havig the data does ot give you ay added value. You keep metioig the cloud. To what extet ca cloud computig help compaies use big data correctly? The cloud does more tha just help compaies use big data. It makes big data possible, eables it i techological terms. I may cases, big data is ot very effective without a dyamic cloud ifrastructure. Nowadays, may compaies are aware of this fact, ad they also uderstad how importat IT is for their busiess. Iformatio is sigificat busiess capital that gives compaies a competitive edge. Both cloud computig ad big data play a essetial role i this. But if usig large amouts of data meas all that extra work, is it worth it i the ed? As I said, hadlig data appropriately is ot a ew cocept that was triggered by big data or cloud computig, so it should t mea all that much extra work. I ay case, it s ot eough to simply keep your techical ifrastructure up-to-date so that you ca have access to all of your data withi a split secod whe you wat to use that iformatio for limited-time marketig campaigs, for example. I-memory computig is essetial if you wat to successfully aggregate these large amouts of data. But the fact is, accordig to a PwC study, that optimal customer satisfactio i the retail sector, for example, ca oly be achieved if compaies are able to make adjustmets to meet customer demad. This icludes delivery times, returs, product selectio ad eve the umber of chaels that customers are able to choose from. Big data aalytics, Iteret resource: 6
7 At the Press of Butto: Better Predictios ad Faster Decisios Whe cars automatically sed o-board diagostic data to a service ceter, optimize fault detectio, exted their lifecycle ad i doig so ehace customer satisfactio a sophisticated big-data strategy is at work. T-Systems Director of Sales Hage Rickma discusses the opportuities that big data creates for compaies. Compaies have bee producig ad aalyzig data for years. So what s so ew about big data? The data quatities ivolved are oly oe part of the picture, albeit a importat oe that affects all idustry sectors without exceptio. Let me offer a example: A major Bavaria automaker reports that its data-geeratio rate has ow reached some 30 gigabytes per day. Ad because cars etwork liks are costatly gettig more powerful, that automaker expects daily data geeratio to reach oe petabyte by the year But while sheer quatity is oe aspect, aother aspect is just as importat for compaies: the eormous diversity of the data ivolved. It will ormally iclude both structured compay data ad ustructured data from social etworks, simulatios ad sesors. What s also ew, however, are the possibilities for aalyzig ad evaluatig such data. Such possibilities lead to better predictios ad faster decisios, to greater efficiecy i product developmet ad to better service. To make use of such possibilities, however, compaies eed access to a complete big-data ecosystem, icludig high-performace storage capacities, a dyamic cloud ifrastructure ad, for example, imemory techologies, which optimize performace i big-data aalysis. Are these treds also chagig the specific opportuities available to compaies? They certaily are. Here s just oe example of how: a commercial real-estate compay i Japa is ow aalyzig data from buildig elevators i real time. This kid of data is really useful. It shows how ofte each floor is beig selected. This ca the be used to determie whe which offices or shops are beig visited less ofte. Such correlatio, i tur, produces treds ad patters that eable the compay to predict whe a commercial teat is goig to give up a particular locatio. For example, a decrease of 60 to 70 percet from the previous moth could poit to a vacacy withi the ext six moths. The real estate compay ca thus act early to fid potetial follow-o teats, ad thereby avoid vacacies ad ca do so behid the scees, as it were, without havig received otice from the curret teat. That s certaily pretty clever. But ca all sectors profit from that sort of data aalysis? The truth is that big-data aalysis is still a blue ocea topic, which meas that idustry is just startig to really discover it. A great deal of research still eeds to be doe, via collaboratio betwee user compaies ad us o the provider side. Hage Rickma Director Sales at T-Systems But we uderstad that although thigs are still at a early stage T-Systems already has umerous solutios to offer for the automobile idustry, for example. Ideed it does. Automakers ca gai a competitive edge by esurig that customers who drive their cars ca always get ay problems solved right away o their first visit to a service ceter. But how ca the cause of a problem be arrowed dow? What similar experieces have which service ceters had with which model? Ad what solutio did they fid? Questios that ca ofte ot be aswered usig the covetioal meas of aalyzig the masses of diagostic data gathered by auto service ceters aroud the world each day. Here is where big data ca make a eormously valuable cotributio. A vehicle ca commuicate proactively with the service ceter ofte solvig a problem before it actually becomes a problem. Very soo, we are goig to be implemetig direct olie liks betwee maufacturers ad service ceters that will make such early fault detectio possible ad, ultimately, greatly ehace customer satisfactio. Maufacturers will be able to avoid expesive recalls, ad they will have much better plaig cotrol over product life cycles. This is all leadig to a wi-wi sceario for maufacturers, service ceters ad customers, which we are goig to make possible by providig a cloud ifrastructure ad high-performace big-data aalysis. I other words, the future of big data lies i productive collaboratio betwee IT compaies ad user compaies? There, ad oly there, is where it lies. We wat to implemet big data i cooperatio with our customers, i a way that miimizes all risks ad highlights the specific beefits. This perspective also applies to legal issues, for example. By o meas do data privacy issues categorically rule out big-data projects or eve make them illegal uder Germa law. At the same time, it s importat to review the legal coformity of ay big-data applicatios from the very start of developmet. We have experts who ca reliably help us avoid the legal pitfalls. At the CeBIT exhibitio, we are presetig seve big-data case studies that illustrate how this works. The examples will give iterested compaies a good impressio of the real possibilities of big data ad show them how secure, powerful big-data ifrastructures ca be set up. I closig, I would like to ivite our customers, ad all iterested parties, to visit us at the Bitkom coferece i mid-april. There as well, we will be offerig iformative presetatios showig how compaies ca use big data to their best advatage. Iteret resource: 7
8 EDITOR: T-Systems Iteratioal GmbH Hahstraße 43d D Frakfurt am Mai Phoe: +49 (0) Iteret: Resposible for cotet: PR & OlieMarketig